US accuses China of industrial-scale AI theft as DeepSeek launches V4 on Huawei chips

US accuses China of "industrial-scale" AI theft. China says it's "slander."

The US is preparing to crack down on China's allegedly "industrial-scale theft of American artificial intelligence labs' intellectual property," the Financial Times reported Thursday. Since the launch of DeepSeek -- a Chinese model that OpenAI claimed was trained using outputs from its models -- other AI firms have accused global rivals of using a method called distillation to steal their IP. In January, Google claimed that "commercially motivated" actors not limited to China attempted to clone its Gemini AI chatbot by promoting the model more than 100,000 times in bids to train cheaper copycats. The next month, Anthropic accused Chinese firms DeepSeek, Moonshot, and MiniMax of using the same tactic to generate "over 16 million exchanges with Claude through approximately 24,000 fraudulent accounts." Also in February, OpenAI confirmed that most attacks it saw originated from China. For the US, these distillation attacks supposedly threaten to help China quickly catch up in the AI race. In a memo that FT reviewed, the director of the White House Office of Science and Technology Policy, Michael Kratsios, warned that "the US government has information indicating that foreign entities, principally based in China, are engaged in deliberate, industrial-scale campaigns to distill US frontier AI systems." According to Kratsios, Chinese campaigns were "leveraging tens of thousands of proxy accounts to evade detection and using jailbreaking techniques to expose proprietary information." His memo said that US firms would soon gain access to government information to help them combat the apparent attacks. So far, AI firms have alleged that such attacks violate their terms of service, but Congress may update laws soon to further equip US companies fighting the alleged fraud. Kratsios confirmed in his memo that the US is exploring measures "to hold foreign actors accountable for industrial-scale distillation campaigns." Congress has already received some marching orders, but it remains unclear how fast lawmakers will act. In an April report, the House's Select Committee on China advised that Congress "should direct the Commerce Department's Bureau of Industry and Security (BIS) and the Department of Justice (DOJ)" to "treat model extraction as industrial espionage" and "impose penalties severe enough to deter Beijing's theft of American innovation."

DeepSeek launches 1.6 trillion parameter V4 on Huawei chips as U.S. escalates AI theft accusations -- U.S. gov't alleges IP theft by DeepSeek and other Chinese AI firms

U.S. State Department warns embassies worldwide about Chinese model distillation. DeepSeek on Friday released a preview of its V4 large language model, the Hangzhou-based startup's most powerful to date, with 1.6 trillion parameters and a 1 million token context window. The model is the first major frontier release optimized for Huawei's Ascend AI processors rather than Nvidia hardware, and it arrived on the same day Reuters reported that the U.S. State Department had sent a diplomatic cable to embassies worldwide instructing staff to warn foreign governments about alleged IP theft by DeepSeek and other Chinese AI firms. V4 comes in two variants: V4-Pro, the flagship, which costs $3.48 per million output tokens, and V4-Flash, a smaller 284 billion parameter version, which costs $0.28. OpenAI currently charges $30 per million output tokens for GPT-5.4, and Anthropic charges $25 for Claude Opus 4.6. DeepSeek, however, acknowledges V4 "falls marginally short" of those closed-source models by roughly three to six months of development, but outperforms every other open-source competitor in agentic coding and reasoning benchmarks. DeepSeek trained its earlier V3 model on 2,048 Nvidia H800 GPUs, and the company has faced multiple investigations over whether it acquired restricted Nvidia hardware through intermediaries in Singapore. V4 sidesteps that supply chain entirely by training on domestic Ascend chips. Huawei confirmed day-zero compatibility across its full Ascend SuperNode product line, including its latest 950 series processors, and DeepSeek said V4-Pro pricing could fall further once Huawei scales up Ascend 950 production in the second half of this year. The diplomatic cable, per Reuters, instructed embassy staff to speak to their foreign counterparts about "concerns over adversaries' extraction and distillation" of U.S. models, naming DeepSeek alongside Moonshot AI and MiniMax. Two days earlier, the White House Office of Science and Technology Policy published a memo accusing Chinese entities of running "deliberate, industrial-scale campaigns" to distill American frontier AI systems. Those accusations build on claims Anthropic made in February, when the company said DeepSeek, Moonshot, and MiniMax had used 24,000 fraudulent accounts to make 16 million exchanges with its Claude model. OpenAI has also accused DeepSeek of distilling its models. China's foreign ministry called the accusations "groundless," according to Reuters, and DeepSeek has previously said its V3 model relied on naturally occurring data collected through web crawling and didn't intentionally use synthetic data generated by OpenAI. The diplomatic cable and the V4 launch both come just weeks before President Trump is scheduled to visit Chinese President Xi Jinping in Beijing for a summit expected to cover semiconductor export controls and IP disputes. Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

US Seeks to Halt US AI Model 'Exploitation' by Chinese Rivals

The Trump administration unveiled measures aimed at preventing Chinese developers from improperly using leading American AI models to build a rival generation of chatbots, marking the first major US response to Silicon Valley companies' complaints that China is piggybacking on their success. In a memo Thursday, the White House Office of Science and Technology Policy said it will promote wider information sharing by US-based developers and increase efforts to help the industry detect unauthorized extraction of their artificial intelligence models. The US government will also work with industry to determine how to rein in such abuses and hold bad actors accountable. "There is nothing innovative about systematically extracting and copying the innovations of American industry, and there is nothing open about supposedly open models that are derived from acts of malicious exploitation," White House Science and Technology Policy Director Michael Kratsios said in the memo. The planned measures represent the most significant US effort so far to rein in a practice known as distillation, where AI developers train systems using results from a parent AI model to create similar capabilities in a new one at a far lower cost. Models made in this way avoid expenses from both research and the costly AI processors needed for original model training. While tolerated for training smaller, less-advanced systems, distillation violates AI companies' terms of use when it's employed to replicate a cutting-edge AI model without permission. The White House clarified in its memo that the US supports a vibrant open-source ecosystem but added that distillation aimed at undermining US research and development investments is unacceptable. The broader effort to crack down on unauthorized distillation seeks to address a growing concern among US companies including OpenAI, Anthropic PBC and Alphabet Inc.'s Google that output from their models is being wrongfully used by Chinese rivals such as DeepSeek, Moonshot and MiniMax to develop products far more cheaply and with fewer safety guardrails. The Office of Science and Technology Policy defines wrongful "industrial-scale" distillation as when foreign entities, primarily based in China, deploy "tens of thousands" of proxy accounts to access leading models and bombard them with queries deliberately aimed at extracting proprietary information that can be used to clone some of the model's capabilities. Though using so-called jail-breaking techniques can result in a nearly-free open-weight Chinese model that mimics a closed-weight US version, the statement warns that unauthorized actors can strip safety protocols through this method, resulting in models that are neither neutral nor truthful. "Foreign entities who build their AI capabilities on such fragile foundations should have little confidence in the integrity and reliability of the models they produce," Kratsios warned in the memo. Accusing Chinese companies of harvesting AI results in an improper fashion represents a provocative move by the administration ahead of next month's highly anticipated summit in Beijing between President Donald Trump and his Chinese counterpart Xi Jinping. The US and China are locked in an all-out competition for global supremacy in artificial intelligence, with each country making the emerging technology a national priority. Top US developers are widely viewed as still being ahead of their Chinese rivals in terms of AI capabilities. Yet at least three US firms have begun to raise the alarm that adversarial distillation poses a risk to their businesses and started sharing information with each other on unauthorized extraction of their models' output. The US government will now join that effort, with a focus on informing companies about the tactics and actors involved. Many models made by Chinese labs are open source and largely free for customers to use. That poses an economic challenge for US AI firms that have kept their systems proprietary, betting that users will pay for access and help offset the hundreds of billions of dollars the firms have spent on data centers and other infrastructure. US officials estimate that illicit extraction of results is costing Silicon Valley labs billions of dollars in annual profit, according to a person familiar with the findings. Scrutiny from US developers over the threat posed by adversarial distillation began after Chinese startup DeepSeek rocked global markets more than a year ago with the release of its breakthrough model. Since then, OpenAI has been investigating whether DeepSeek improperly extracted results from leading American models to develop the R1 chatbot. Anthropic and Google DeepMind have voiced similar concerns, and earlier this year, all three labs posted publicly that they had detected their models had been improperly queried and copied through large-scale data requests on their platforms. Absent clearer disclosures from China, it's difficult to parse how much of China's AI model success is the result of its own innovation or simply copying US technology.

Exclusive: US State Dept orders global warning about alleged China AI thefts by DeepSeek, others

WASHINGTON, April 24 (Reuters) - The U.S. State Department has ordered a global push to bring attention to what it says are widespread efforts by Chinese companies, including AI startup DeepSeek, to steal intellectual property from U.S. artificial intelligence labs, according to a diplomatic cable seen by Reuters. The cable said its purpose was to "warn of the risks of utilizing AI models distilled from U.S. proprietary AI models, and lay the groundwork for potential follow-up and outreach by the U.S. government." Distillation is the process of training smaller AI models using output from larger, more expensive ones as part of an effort ⁠to lower the costs of training a powerful new AI tool. DeepSeek, the Chinese startup whose low-cost AI model stunned the world last year, on Friday launched a preview of a highly anticipated new model adapted for Huawei chip technology, underlining China's growing autonomy in the sector. The State Department, DeepSeek and the Chinese Embassy in Washington did not immediately respond to a request for comment. The cable also mentioned Chinese AI firms Moonshot AI and MiniMax. Neither company immediately responded to a request for comment. This week, the White House made similar accusations, which the Chinese Embassy in Washington called "baseless allegations," adding that Beijing "attaches great importance to the protection of intellectual property rights." The cable, dated Friday and sent to diplomatic and consular posts around the world, instructs diplomatic staff to speak to their ⁠foreign counterparts about "concerns over adversaries' extraction and distillation of U.S. A.I. models." "A separate demarche request and message has been sent to Beijing for raising with China," the document states. The cable, which has not been previously reported, signals the Trump administration is taking seriously the growing concerns about Chinese distillation of U.S. AI models. "AI models developed from surreptitious, unauthorized distillation campaigns enable foreign actors to release products that appear ⁠to perform comparably on select benchmarks at a fraction of the cost but do not replicate the full performance of the original system," the cable said, adding that the campaigns also "deliberately strip security protocols from the resulting models and undo mechanisms that ensure those AI models ⁠are ideologically neutral and truth‑seeking." OpenAI warned U.S. lawmakers that Chinese artificial intelligence startup DeepSeek was targeting the ChatGPT maker and the nation's leading AI companies to replicate models and use them for its own training, Reuters reported in February. The memo and follow-up ⁠cable, released just weeks before U.S. President Donald Trump is set to visit Chinese President Xi Jinping in Beijing, promise to raise tensions in a long-running tech war between the rival superpowers, which had been lowered by a detente brokered last October. Reporting by Raphael Satter and Alexandra Alper in Washington; editing by Chris Sanders and David Gregorio Our Standards: The Thomson Reuters Trust Principles., opens new tab * Suggested Topics: * Artificial Intelligence Raphael Satter Thomson Reuters Reporter covering cybersecurity, surveillance, and disinformation for Reuters. Work has included investigations into state-sponsored espionage, deepfake-driven propaganda, and mercenary hacking.

White House accuses China of 'industrial scale' theft of AI technology

The White House has accused China of undertaking industrial-scale theft of American artificial intelligence labs' intellectual property and warned that it would crack down on a practice that exploits US innovation. "The US government has information indicating that foreign entities, principally based in China, are engaged in deliberate, industrial-scale campaigns to distil US frontier AI systems," Michael Kratsios, director of the White House Office of Science and Technology Policy, wrote in a memo seen by the FT. The accusation marks the latest escalation in tensions around Chinese groups allegedly raiding advanced American AI research amid an arms race to lead in the technology. It comes just weeks before President Donald Trump will meet President Xi Jinping in Beijing. The issue gained attention after China's DeepSeek was accused of using distillation -- the process of training smaller AI models based on the output of larger ones -- to build a powerful product at a lower cost. Kratsios' memo to government departments said the administration would share information with American AI companies about "attempts by foreign actors to conduct unauthorised, industrial-scale distillation" and help them co-ordinate against attacks. He said Chinese campaigns were "leveraging tens of thousands of proxy accounts to evade detection and using jailbreaking techniques to expose proprietary information". The US would explore measures "to hold foreign actors accountable for industrial-scale distillation campaigns", Kratsios added. Chris McGuire, a technology security expert at the Council on Foreign Relations said: "Chinese AI firms are relying on distillation attacks to offset deficits in AI computing power and illicitly reproduce the core capabilities of US models." McGuire said the US should ban Chinese groups from accessing US models and sanction entities that conduct or enable distillation, as well as tighten export controls to prevent China from smuggling or remotely accessing US AI chips. The Chinese embassy did not respond to a request for comment. US AI firms, including Anthropic and OpenAI, have increasingly voiced concern about distillation by Chinese groups, which they argue enables foreign labs to close the competitive advantage that the US enjoys because of export controls on advanced American chips. Kratsios said distillation was a vital part of the AI ecosystem when used legitimately to make lighter-weight models but "industrial distillation" used to undermine American research and development was "unacceptable". He added that while models created by "surreptitious, unauthorised distillation campaigns" did not match the performance of the original models, they can benefit foreign groups because of the significantly lower cost. In February, Anthropic accused three leading Chinese AI companies -- DeepSeek, Moonshot and MiniMax -- of distillation attacks on its models. That came after OpenAI said in early 2025 that it had evidence that DeepSeek had used outputs from its GPT models to train its model in violation of its terms of service. American AI companies are concerned that distilled models pose national security risks because they lack the safeguards that, for example, prevent the development of bioweapons or malicious cyber attacks. The House Foreign Affairs Committee on Wednesday passed a slew of bills designed to make it harder for China to catch up with the US in the AI race. One bill tackles distillation by requiring the administration to consider adding groups that employ it to the "entity list" -- an export blacklist that would make it very hard for US companies to sell technology to the groups.

U.S. State Department orders global warning about alleged China AI thefts by DeepSeek, others: Reuters

The U.S. State Department has ordered a global push to bring attention to what it says are widespread efforts by Chinese companies, including AI startup DeepSeek, to steal intellectual property from U.S. artificial intelligence labs, according to a diplomatic cable seen by Reuters. The cable said its purpose was to "warn of the risks of utilizing AI models distilled from U.S. proprietary AI models, and lay the groundwork for potential follow-up and outreach by the U.S. government." Distillation is the process of training smaller AI models using output from larger, more expensive ones as part of an effort to lower the costs of training a powerful new AI tool. DeepSeek, the Chinese startup whose low-cost AI model stunned the world last year, on Friday launched a preview of a highly anticipated new model adapted for Huawei chip technology, underlining China's growing autonomy in the sector. The State Department, DeepSeek and the Chinese Embassy in Washington did not immediately respond to a request for comment. The cable also mentioned Chinese AI firms Moonshot AI and MiniMax. Neither company immediately responded to a request for comment. This week, the White House made similar accusations, which the Chinese Embassy in Washington called "baseless allegations," adding that Beijing "attaches great importance to the protection of intellectual property rights." The cable, dated Friday and sent to diplomatic and consular posts around the world, instructs diplomatic staff to speak to their foreign counterparts about "concerns over adversaries' extraction and distillation of U.S. A.I. models." "A separate demarche request and message has been sent to Beijing for raising with China," the document states. The cable, which has not been previously reported, signals the Trump administration is taking seriously the growing concerns about Chinese distillation of U.S. AI models. "AI models developed from surreptitious, unauthorized distillation campaigns enable foreign actors to release products that appear to perform comparably on select benchmarks at a fraction of the cost but do not replicate the full performance of the original system," the cable said, adding that the campaigns also "deliberately strip security protocols from the resulting models and undo mechanisms that ensure those AI models are ideologically neutral and truth-seeking." OpenAI warned U.S. lawmakers that Chinese artificial intelligence startup DeepSeek was targeting the ChatGPT maker and the nation's leading AI companies to replicate models and use them for its own training, Reuters reported in February. The memo and follow-up cable, released just ?weeks ?before U.S. President Donald Trump is set to visit Chinese President Xi Jinping in Beijing, promise to raise tensions in a long-running tech war between the rival superpowers, which had been lowered by a detente brokered last October.

Trump administration vows crackdown on Chinese companies 'exploiting' AI models made in US

WASHINGTON (AP) -- The Trump administration is vowing to crack down on foreign tech companies' exploitation of U.S. artificial intelligence models, singling out China at a time that country is narrowing the gap with the U.S. in the AI race. In a Thursday memo, Michael Kratsios, the president's chief science and technology adviser, accused foreign entities "principally based in China" of engaging in deliberate, industrial-scale campaigns to "distill," or extract capabilities from, leading AI systems made in the U.S. and "exploiting American expertise and innovation." The administration, Kratsios wrote, will work with American AI companies to identify such activities, build defenses and find ways to punish offenders. The memo arrives at a time when China is challenging U.S. dominance in artificial intelligence, an area where the White House says the U.S. must prevail to set global standards and reap economic and military benefits. But the U.S.-China gap in performance of top AI models has "effectively closed," according to a recent report from Stanford University's Institute for Human-Centered AI. China's embassy in Washington said it opposed "the unjustified suppression of Chinese companies by the U.S." "China has always been committed to promoting scientific and technological progress through cooperation and healthy competition. China attaches great importance to the protection of intellectual property rights," said Liu Pengyu, the embassy spokesperson. Kratsios' memo also came the same week that the House Foreign Affairs Committee offered unanimous, bipartisan support for a bill to set up a process to identify foreign actors that extract "key technical features" of closed-source, U.S.-owned AI models and to punish them with measures including sanctions. "Model extraction attacks are the latest frontier of Chinese economic coercion and theft of U.S. intellectual property," said Rep. Bill Huizenga, R-Mich., who sponsored the bill. "American AI models are demonstrating transformative cyber capabilities, and it is critical we prevent China from stealing these technological advancements." Last year, the Chinese start-up DeepSeek rattled U.S. markets when it released a large language model that could compete with U.S. AI giants but at a fraction of the cost. David Sacks, then serving as President Donald Trump's AI and crypto adviser, suggested that DeepSeek copied U.S. models. "There's substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI's models," Sacks said then. In a February letter to U.S. lawmakers, OpenAI, the developer of ChatGPT, made similar allegations and said China should not be allowed to advance "autocratic AI" by "appropriating and repackaging American innovation." Anthropic, the maker of the Claude chatbot, in February accused DeepSeek and two other China-based AI laboratories of engaging in campaigns to "illicitly extract Claude's capabilities to improve their own models" using the distillation technique that "involves training a less capable model on the outputs of a stronger one." Anthropic said distillation can be a legitimate way to train AI systems but it's a problem when competitors "use it to acquire powerful capabilities from other labs in a fraction of the time, and at a fraction of the cost, that it would take to develop them independently." But it can go both ways. San Francisco-based startup Anysphere, maker of the popular coding tool Cursor, recently acknowledged that its latest product was based on an open-source model made by Chinese company Moonshot AI, maker of the chatbot Kimi. Kyle Chan, a fellow at the Washington-based think tank The Brookings Institution and an expert on China's technology development, said it will be like "looking for needles in an enormous haystack" to separate unauthorized distillation from the vast volume of legitimate requests for data. But information sharing and coordination among U.S. AI labs could help, and the federal government can play an important role in facilitating anti-distillation efforts across labs, Chan said. It's hard to assess how far the House bill can go, but Chan said Trump may not want to rock the boat with Chinese President Xi Jinping ahead of a planned mid-May state visit to Beijing. ___ AP Technology Writer Matt O'Brien contributed to this report from Providence, Rhode Island.

White House memo claims mass AI theft by Chinese firms

The White House has said it will work more closely with US artificial intelligence (AI) firms to combat "industrial-scale campaigns" by foreign actors to steal advances in the technology. Michael Kratsios, Director of Science and Technology Policy, wrote in an internal memo that the administration had new information indicating "foreign entities, principally based in China" were exploiting American firms. Through a process called "distilling", such firms are essentially copying AI technology developed by US companies, he said. A representative of China's US embassy in Washington DC said its development was "the result of its own dedication and effort as well as international cooperation". In the memo, Kratsios said the aim was to "systematically undermine American research and development and access proprietary information". In an attempt avoid and halt "malicious exploitation," he said the White House will be doing four things: * sharing more information with US AI companies about "tactics employed and actors involved" in distillation campaigns * working to "better coordinate" with companies to fight the attacks * develop a set of "best practices to identify, mitigate, and remediate" them * "explore" how the White House can hold foreign actors accountable for such distillation The memo did not detail any specific plans for action against foreign entities found to be undertaking distillation of US AI technology. A White House spokesperson declined to comment beyond the memo. A representative of China's US embassy in Washington DC took issue with "the unjustified suppression of Chinese companies by the US" in response to the memo "China is not only the world's factory but is also becoming the world's innovation lab," the representative added. "China's development is the result of its own dedication and effort as well as international cooperation that delivers mutual benefits." Distillation campaigns are carried out by firms that usually operate many thousands of individual accounts for a given AI chatbot or tool, allowing them to appear as normal users. Those accounts then undertake more coordinated attempts to "jailbreak" or otherwise expose information about AI models that is not supposed to be made public, which is saved and applied to their own AI model building and training. "As methods to detect and mitigate industrial-scale distillation grow more sophisticated, foreign entities who build their AI capabilities on such fragile foundations should have little confidence in the integrity and reliability of the models they produce," Kratsios said.

The US just told China to stop copying its AI. Enforcing that is the hard part.

Summary: The White House OSTP released a policy memo accusing China of "industrial-scale" distillation of US AI models, committing to share intelligence with US AI companies and explore accountability measures. OpenAI accused DeepSeek of distilling its models in February; Anthropic named DeepSeek, MiniMax, and Moonshot AI as having created 24,000 fraudulent accounts generating 16+ million exchanges with Claude. The Deterring American AI Model Theft Act (H.R. 8283) was introduced on 15 April. The memo arrives three weeks before a planned Trump-Xi summit on 14 May. The White House accused China on Wednesday of conducting "industrial-scale" theft of American artificial intelligence, releasing a policy memorandum that commits the government to sharing intelligence with US AI companies about foreign distillation campaigns and exploring measures to hold the perpetrators accountable. Michael Kratsios, director of the Office of Science and Technology Policy, said the US "has evidence that foreign entities, primarily in China, are running industrial-scale distillation campaigns to steal American AI. We will be taking action to protect American innovation." The memo lands three weeks before a planned Trump-Xi summit in Beijing on 14 May, positioning AI technology protection as both a national security imperative and a negotiating chip. Distillation is the technique at the centre of the dispute. It does not require stealing model weights or breaking into servers. A distiller feeds thousands or millions of carefully constructed queries to a frontier AI model, collects the responses, and uses those responses to train a cheaper rival model that approximates the original's capabilities at a fraction of the cost. It is, in effect, learning from the teacher's answers rather than the teacher's brain. The legal status of this technique is unsettled. The strategic implications are not. The OSTP memo builds on allegations that US AI companies have been making since February. OpenAI sent a formal memo to the House Select Committee on China on 12 February accusing DeepSeek of distilling its models. OpenAI said it had identified accounts associated with DeepSeek employees that developed methods to circumvent access restrictions, routing queries through obfuscated third-party proxies to extract outputs at scale. OpenAI's terms of service explicitly prohibit using outputs to create "imitation frontier AI models." DeepSeek has not publicly responded to the allegations. Anthropic published more detailed evidence on 23 February, naming three Chinese laboratories. DeepSeek, it said, conducted more than 150,000 exchanges with Claude focused on foundational logic and alignment techniques. MiniMax drove the most traffic, with more than 13 million exchanges. Moonshot AI generated more than 3.4 million exchanges targeting agentic reasoning, tool use, coding, and computer vision. Across the three firms, Anthropic identified approximately 24,000 fraudulent accounts that generated more than 16 million exchanges with Claude. The accounts used jailbreaking techniques to expose proprietary information and circumvented geofencing through commercial proxy services. By early April, OpenAI, Anthropic, and Google had begun sharing distillation threat intelligence through the Frontier Model Forum, a coalition originally founded in 2023 with Microsoft. The arrangement is modelled on cybersecurity threat-sharing frameworks: when one company detects an attack pattern, it flags it for the others. That three fierce competitors agreed to cooperate on anything is itself a measure of how seriously they take the threat. DeepSeek proved that frontier AI performance no longer requires Silicon Valley-scale resources, and the question the US government is now asking is how much of that efficiency was earned and how much was extracted. The OSTP memo is a policy statement, not an executive order or a binding regulation. It directs federal departments to share intelligence with US AI developers about foreign distillation attempts, help industry strengthen technical defences, and explore accountability measures for foreign actors. No specific sanctions, entity list additions, or enforcement actions were announced on Wednesday. The memo's practical force will depend on what follows it. Congress is moving in parallel. On 15 April, Representative Bill Huizenga introduced the Deterring American AI Model Theft Act of 2026, co-sponsored by Representative John Moolenaar, who chairs the House Select Committee on China. The bill would direct the government to identify entities using "improper query-and-copy techniques" and impose sanctions through the Commerce Department blacklist. The House Select Committee held a hearing on 16 April titled "China's Campaign to Steal America's AI Edge," with witnesses from Brookings, the Silverado Policy Accelerator, and the America First Policy Institute. The issue has bipartisan support. Roll Call reported that "winning the AI arms race holds appeal for both parties." The legal theory underpinning prosecution remains unclear. The Protecting American Intellectual Property Act, signed in January 2023, authorises sanctions for trade secret theft, but whether extracted model outputs qualify as trade secrets under existing frameworks is an open question. The South China Morning Post noted that Anthropic's distillation charges "expose an AI training grey area," and legal analysts at Just Security have argued that the case for imposing costs on distillation requires targeted government intervention precisely because existing intellectual property law does not cleanly cover it. The shift from hardware controls to model-level protections represents an acknowledgement that the first line of defence is leaking. The US has been restricting China's access to advanced AI chips since October 2022, broadening the rules in October 2023 and again with the AI Diffusion Rule in January 2025. In January 2026, the Bureau of Industry and Security shifted its review of H200 and AMD MI325X exports to China from a presumption of denial to case-by-case review, while the White House simultaneously imposed a 25% tariff on advanced semiconductors. Nvidia was permitted to sell its H20 inference chip; AMD its MI308. But hardware controls are circumvented in practice. A $2.5 billion scheme to smuggle Nvidia AI chips to China through Super Micro's co-founder was charged in March. Jensen Huang warned that DeepSeek optimising for Huawei chips would be a "horrible outcome" for America, because it would eliminate the hardware chokepoint entirely. If advanced chips can be smuggled despite export controls, and if Chinese chipmakers are closing the gap with domestic alternatives, then preventing access to the models themselves becomes the critical second layer of the technology denial strategy. Proposals to tag AI chips with unique identifiers represent a third layer, tracking hardware flows to prevent diversion. The emerging architecture is defence in depth: control the chips, control the models, and track both. Distillation is not the only channel through which US AI technology reaches Chinese laboratories. Meta's Llama models are open source, meaning the weights are publicly available for download. Chinese researchers from PLA-linked institutions fine-tuned Llama 13B on military data to create ChatBIT, a model designed for military intelligence applications. Meta's acceptable use policy prohibits military and espionage applications, but the company has no technical means to enforce that restriction on open-source releases. Once the weights are published, control is relinquished. Meta responded by opening Llama to the US military and Five Eyes allies while maintaining the ban for adversaries, a policy distinction that is legally meaningful and practically unenforceable. The tension between open-source AI and national security has been building for years but has not produced a coherent policy resolution. Open-source models drive research, attract talent, and create ecosystems that benefit American companies. Restricting them would slow US innovation while pushing Chinese developers toward domestic alternatives. Not restricting them means providing the foundational technology for adversary military applications. The Huizenga bill focuses on distillation, the unauthorised extraction of capability from closed models, rather than on open-source distribution, sidestepping the harder question. The US-China chip war has already drawn allies into the effort, with the Netherlands restricting ASML's lithography exports under American pressure. Model-level restrictions would require a different enforcement architecture. Chips are physical objects that cross borders. Distillation happens over the internet, through API calls that can be routed through any jurisdiction. Detecting it requires the kind of behavioural analysis that Anthropic performed when it identified 24,000 fraudulent accounts, not the kind of customs inspection that catches smuggled hardware. The Trump-Xi summit on 14 May will test whether the OSTP memo is the beginning of a sustained enforcement campaign or a negotiating position designed to extract concessions. China wants the US to loosen technology controls, remove more than 1,000 Chinese firms from entity lists, and reduce investment restrictions. The US wants China to stop distilling its AI models, stop smuggling its chips, and stop fine-tuning its open-source models for military use. The gap between those positions is wide enough that neither side is likely to get what it wants. What the memo establishes, regardless of the summit's outcome, is that the US now treats AI model protection as a category of national security alongside chip export controls and semiconductor equipment restrictions. The question is no longer whether distillation is a problem. It is whether the government can enforce a border around something that has no physical form.

Trump Admin Accuses China of â€~Industrial-Scale’ Theft of AI Tech. What Does That Even Mean?

The Trump administration (which has argued that scraping an infinite amount of copyright-protected data for training AI models should be considered fair use) is accusing China of "industrial-scale" theft of intellectual property from American AI companies and is threatening to crack down on actors accused of doing the jacking. According to the Financial Times, the accusations of theft were apparently levied in a memo from Michael Kratsios, the director of the White House Office of Science and Technology Policy, and distributed to multiple government agencies. It directed those working in the departments to share information with AI companies to make them aware of any attempts by foreign actors to access their sensitive information. At issue is the apparent practice of distillation, in which a larger model’s outputs are used to train a smaller one, enabling the smaller model to mimic some of the larger model’s performance more cheaply and with less computational overhead. It's the practice that Anthropic accused several China-based AI labs of doing earlier this year. OpenAI also claimed that DeepSeek used distillation techniques to train its open-source model, arguing that the lab was trying to "free-ride" on the back of work done by US firms. Pot, have you met kettle? Most of these companies have been very protective of stuff in the black box of AI, the weights and data that make their models tick. But we know much of that material is (or was at one point) protected by copyright. Dubious practices, like Anthropic buying, scanning, and destroying millions of books, give tech giants potential legal cover as courts continue to debate whether or not AI model training constitutes fair use, but the argument that something transformational is happening under the hood gets complicated when models regularly reproduce training material nearly verbatim. Under that context, it's interesting to see how far these companies will go to protect their own material. When Anthropic recently saw the source code for its Claude Code product leak online, the company reportedly issued thousands of copyright takedown requests to prevent people from republishing and sharing it. According to the New York Times, there was one notable exception to the widespread removal attempt: a rewritten version hosted on GitHub. One quick-thinking leaker used AI agents to translate the leak into another programming language, which was apparently enough for Anthropic to deem it transformative and outside its claim of ownership. Given that the Copyright Office has determined that works created by AI systems without human input are not eligible for copyright protections, and companies like Anthropic like to brag about how much of their code is now completed by autonomous AI agents, it's increasingly interesting to learn what is and isn't covered. But it does seem that AI companies believe they should get special treatment. Copyright doesn't apply when they want to bypass it, but it's an essential protection when it means someone might get their hands on their work.

White House accuses China of 'industrial scale' theft of AI technology, FT reports

April 23 (Reuters) - The White House has accused China of undertaking industrial-scale theft of U.S. artificial intelligence labs' intellectual property and ⁠warned it would crack down on a practice that exploits U.S. innovation, the Financial Times reported on Thursday. The ⁠report cited a memo written by Michael Kratsios, director ⁠of the White House Office of Science ⁠and Technology Policy. Reuters could not ⁠immediately verify the report. Reporting by Ruchika Khanna in Bengaluru Our Standards: The Thomson Reuters Trust Principles., opens new tab

White House: China using proxies to copy US frontier AI models

The White House on Thursday accused China-based actors of targeting U.S. artificial intelligence labs in coordinated campaigns to extract intellectual property. The warning comes just weeks before U.S. President Donald Trump is expected to meet Chinese President Xi Jinping in Beijing, raising concerns that the issue could overshadow upcoming talks. Michael Kratsios, director of the White House Office of Science and Technology Policy, detailed the allegations in a memo circulated to federal agencies.

White House warns of 'industrial-scale' efforts in China to rip off U.S. AI tech

White House accuses China of 'industrial-scale' campaigns to 'distill U.S. frontier AI systems' The Trump administration on Thursday accused Chinese entities of waging "industrial-scale campaigns" to rip off U.S. artificial intelligence systems, and said it will explore ways to hold the foreign actors accountable. "There is nothing innovative about systematically extracting and copying the innovations of American industry," Michael Kratsios, the top science and technology advisor to President Donald Trump, said in a memo on alleged Chinese "distillation" operations to train smaller AI models off of larger ones. The U.S. government has previously accused China of targeting American AI technology and intellectual property. Kratsios warned that as it gets easier to detect and prevent large-scale "distillation" operations, the entities that "build their AI capabilities on such fragile foundations" should lose confidence in "the integrity and reliability of the models they produce." U.S. information indicates that the campaigns to "distill" U.S. frontier AI systems are coming from mostly China-based entities, he said. The efforts involve using tens of thousands of proxy accounts and jailbreaking techniques to secretly "expose proprietary information," Kratsios said. Attempts to copy U.S. models through "surreptitious, unauthorized distillation campaigns" won't result in AI systems with the same performance as the originals, the advisor noted. But they "enable foreign actors to release products that appear to perform comparably on select benchmarks at a fraction of the cost," he said.

Trump administration vows crackdown on Chinese firms 'exploiting' U.S. AI models

WASHINGTON -- The Trump administration is vowing to crack down on foreign tech companies' exploitation of U.S. artificial intelligence models, singling out China at a time that country is narrowing the gap with the U.S. in the AI race. In a Thursday memo, Michael Kratsios, the president's chief science and technology adviser, accused foreign entities "principally based in China" of engaging in deliberate, industrial-scale campaigns to "distill," or extract capabilities from, leading AI systems made in the U.S. and "exploiting American expertise and innovation." The administration, Kratsios wrote, will work with American AI companies to identify such activities, build defenses and find ways to punish offenders. The memo arrives at a time when China is challenging U.S. dominance in artificial intelligence, an area where the White House says the U.S. must prevail to set global standards and reap economic and military benefits. But the U.S.-China gap in performance of top AI models has "effectively closed," according to a recent report from Stanford University's Institute for Human-Centered AI. China's embassy in Washington said it opposed "the unjustified suppression of Chinese companies by the U.S." "China has always been committed to promoting scientific and technological progress through cooperation and healthy competition. China attaches great importance to the protection of intellectual property rights," said Liu Pengyu, the embassy spokesperson. Kratsios' memo also came the same week that the House Foreign Affairs Committee offered unanimous, bipartisan support for a bill to set up a process to identify foreign actors that extract "key technical features" of closed-source, U.S.-owned AI models and to punish them with measures including sanctions. "Model extraction attacks are the latest frontier of Chinese economic coercion and theft of U.S. intellectual property," said Rep. Bill Huizenga, R-Mich., who sponsored the bill. "American AI models are demonstrating transformative cyber capabilities, and it is critical we prevent China from stealing these technological advancements." Last year, the Chinese start-up DeepSeek rattled U.S. markets when it released a large language model that could compete with U.S. AI giants but at a fraction of the cost. David Sacks, then serving as President Donald Trump's AI and crypto adviser, suggested that DeepSeek copied U.S. models. "There's substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI's models," Sacks said then. In a February letter to U.S. lawmakers, OpenAI, the developer of ChatGPT, made similar allegations and said China should not be allowed to advance "autocratic AI" by "appropriating and repackaging American innovation." Anthropic, the maker of the Claude chatbot, in February accused DeepSeek and two other China-based AI laboratories of engaging in campaigns to "illicitly extract Claude's capabilities to improve their own models" using the distillation technique that "involves training a less capable model on the outputs of a stronger one." Anthropic said distillation can be a legitimate way to train AI systems but it's a problem when competitors "use it to acquire powerful capabilities from other labs in a fraction of the time, and at a fraction of the cost, that it would take to develop them independently." But it can go both ways. San Francisco-based startup Anysphere, maker of the popular coding tool Cursor, recently acknowledged that its latest product was based on an open-source model made by Chinese company Moonshot AI, maker of the chatbot Kimi. Kyle Chan, a fellow at the Washington-based think tank The Brookings Institution and an expert on China's technology development, said it will be like "looking for needles in an enormous haystack" to separate unauthorized distillation from the vast volume of legitimate requests for data. But information sharing and coordination among U.S. AI labs could help, and the federal government can play an important role in facilitating anti-distillation efforts across labs, Chan said. It's hard to assess how far the House bill can go, but Chan said Trump may not want to rock the boat with Chinese President Xi Jinping ahead of a planned mid-May state visit to Beijing.

U.S. accuses China of "industrial-scale" campaigns to steal AI secrets

Why it matters: The accusation pushes the U.S.-China AI rivalry into more confrontational territory -- and could complicate President Trump's upcoming visit to Beijing. Driving the news: Michael Kratsios, director of the White House Office of Science and Technology Policy, sent a memo Thursday to federal agency heads accusing mostly China-based actors of using proxy accounts to evade detection and jailbreak models to "expose proprietary information" and "extract capabilities from American AI models." * Distillation attacks involve querying proprietary models, like Claude or Gemini, millions of times via APIs to build datasets that replicate how the systems behave. * Kratsios said these campaigns enable foreign actors to release models that appear to match U.S. AI capabilities at a fraction of the cost. * He added that such tactics can also strip away guardrails meant to keep outputs "ideologically neutral and truth-seeking." State of play: The warning comes as Trump prepares for a highly anticipated trip next month to Beijing, where he's expected to push for economic concessions and reset parts of the U.S.-China relationship. * OpenAI and Anthropic both said earlier this year that China-based firms -- including DeepSeek, Moonshot AI and MiniMax -- were behind wide-scale distillation attacks on their models. The big picture: The U.S. has long accused China of stealing intellectual property from American companies as part of broader cyber espionage efforts. * In 2024, the Justice Department indicted a former Google software engineer for stealing AI trade secrets and sharing them with two Chinese companies. Yes, but: Kratsios argued the abilities of these distilled models may not hold up over time. * "As methods to detect and mitigate industrial-scale distillation grow more sophisticated, foreign entities who build their AI capabilities on such fragile foundations should have little confidence in their integrity and reliability of the models they produce," he wrote. What's next: Kratsios said the Trump administration plans to share intelligence with U.S. AI companies on these campaigns, including the tactics they used, and help the private sector develop defenses.

White House Accuses China of 'Industrial-Scale' Theft From American AI Models - Decrypt

The memo calls for federal agencies and private industry to develop defenses and hold foreign actors accountable. The White House warned Thursday that "foreign entities" are allegedly carrying out "industrial-scale" campaigns to copy the capabilities of American-based artificial intelligence models, using tactics including jailbreaking and networks of fake accounts to extract proprietary information and replicate their performance. In a memorandum titled "Adversarial Distillation of American AI Models," Assistant to the President for Science and Technology and Director of the Office of Science and Technology Policy Michael Kratsios said the U.S. government has information indicating coordinated efforts to distill U.S. frontier AI systems. "The U.S. has evidence that foreign entities, primarily in China, are running industrial-scale distillation campaigns to steal American AI," Kratsios wrote on X. "We will be taking action to protect American innovation." According to President Donald Trump's administration, the campaigns are using "tens of thousands of proxy accounts" to evade detection and exploit jailbreak techniques to systematically extract capabilities, in what is known as a distillation attack. A distillation attack is a method of training a smaller AI model to learn from the outputs of a larger one. The issue has become a growing concern among U.S. AI companies. In February, Anthropic accused Chinese AI labs DeepSeek, Moonshot, and MiniMax of extracting millions of Claude responses -- using roughly 24,000 fraudulent accounts -- to train competing systems. Models developed through unauthorized distillation campaigns may not match the full performance of the originals. Still, they can appear comparable on select benchmarks at a fraction of the cost. The administration warned that distillation attacks could also remove security safeguards and other controls designed to keep AI systems "ideologically neutral and truth-seeking." The Trump administration said federal agencies will work with U.S. AI companies to strengthen protections around frontier models, coordinate with private industry to develop defenses against large-scale distillation campaigns, and explore ways to hold foreign actors accountable. While the memo acknowledged that lawful distillation can help create smaller, more efficient open-source and open-weight models, it said unauthorized efforts to copy American AI systems cross the line. "There is nothing innovative about systematically extracting and copying the innovations of American industry," the memo said. The Office of Science and Technology did not immediately respond to a request for comment by Decrypt.

White House's claim that China is 'engaged in deliberate, industrial-scale campaigns to distil US frontier AI systems' called 'pure slander' by Chinese embassy

A memo seen by the Financial Times places the blame on "foreign entities, principally based in China." It looks like the US government might be getting a little more serious about tackling the growth and development of foreign AI companies, judging by a White House memo seen by the Financial Times. The memo reportedly says: "The US government has information indicating that foreign entities, principally based in China, are engaged in deliberate, industrial-scale campaigns to distil US frontier AI systems." The White House will apparently share information about all this with other American AI companies and help them coordinate against these operations. Distillation, in the context of AI, is the process of training a model based on the outputs of a more advanced one, so it can learn how to replicate it. The Financial Times reported earlier in the year that Anthropic had accused DeepSeek, Moonshot, and MiniMax of "industrial-scale distillation attacks on our models." Given Anthropic has been in more talks with the White House this month to discuss working together, it's not a big stretch to assume that the company is at least in part responsible for bringing the White House's attention to this distillation issue -- or at least to its severity. It's not only distillation that is an issue, though, apparently, as Kraitsios reportedly said that these Chinese operations were "leveraging tens of thousands of proxy accounts to evade detection and using jailbreaking techniques to expose proprietary information." The US is said to be exploring measures to hold foreign actors to account. In response, the Chinese embassy in Washington has said these accusations are "pure slander" and that "China attaches great importance to the protection of intellectual property rights." Michael Kratsios has been Trump's top science and tech advisor for some time and also serves on the President's Council of Advisors on Science and Technology (PCAST). This group was recently reestablished with a bunch of business-oriented members including Meta's Mark Zuckerberg and AMD's Lisa Su. Assuming it's legit, the memo will certainly have some weight behind it -- at least to the President's eyes and ears given its origin. Such concerns will likely only become more pronounced as things go forward, given DeepSeek 4 just launched and is supposedly offering "cost-effective 1M context length" -- ie, cheap AI with better conversational memory. It's something I'm sure no American AI company is sleeping on, so these companies and the US will be looking to shore up any holes that might allow corporate copying.

Trump administration vows crackdown on China's 'exploiting' of AI models made in the U.S.

Michael Kratsios, the president's technology adviser, accused Chinese companies of stealing leading AI models made in the U.S. The Trump administration is vowing to crack down on foreign tech companies' exploitation of U.S. artificial intelligence models, singling out China at a time that country is narrowing the gap with the U.S. in the AI race. In a Thursday memo, Michael Kratsios, the president's chief science and technology adviser, accused foreign entities "principally based in China" of engaging in deliberate, industrial-scale campaigns to "distill," or extract capabilities from, leading AI systems made in the U.S. and "exploiting American expertise and innovation." The administration, Kratsios wrote, will work with American AI companies to identify such activities, build defenses and find ways to punish offenders. The memo arrives at a time when China is challenging U.S. dominance in artificial intelligence, an area where the White House says the U.S. must prevail to set global standards and reap economic and military benefits. But the U.S.-China gap in performance of top AI models has "effectively closed," according to a recent report from Stanford University's Institute for Human-Centered AI. China's embassy in Washington said it opposed "the unjustified suppression of Chinese companies by the U.S." "China has always been committed to promoting scientific and technological progress through cooperation and healthy competition. China attaches great importance to the protection of intellectual property rights," said Liu Pengyu, the embassy spokesperson. In Beijing, China's Foreign Ministry spokesperson Guo Jiakun told reporters Friday that the U.S. claims are groundless and were smearing the achievements of China's artificial intelligence industry. "China firmly opposes this. We urge the U.S. to respect facts, discard prejudice, stop suppressing China's technological development, and do more to promote scientific and technological exchange and cooperation between the two countries," he said. Kratsios' memo also came the same week that the House Foreign Affairs Committee offered unanimous, bipartisan support for a bill to set up a process to identify foreign actors that extract "key technical features" of closed-source, U.S.-owned AI models and to punish them with measures including sanctions. "Model extraction attacks are the latest frontier of Chinese economic coercion and theft of U.S. intellectual property," said Rep. Bill Huizenga, R-Mich., who sponsored the bill. "American AI models are demonstrating transformative cyber capabilities, and it is critical we prevent China from stealing these technological advancements." Last year, the Chinese startup DeepSeek rattled U.S. markets when it released a large language model that could compete with U.S. AI giants but at a fraction of the cost. David Sacks, then serving as President Donald Trump's AI and crypto adviser, suggested that DeepSeek copied U.S. models. "There's substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI's models," Sacks said then. In a February letter to U.S. lawmakers, OpenAI, the developer of ChatGPT, made similar allegations and said China should not be allowed to advance "autocratic AI" by "appropriating and repackaging American innovation." Anthropic, the maker of the Claude chatbot, in February accused DeepSeek and two other China-based AI laboratories of engaging in campaigns to "illicitly extract Claude's capabilities to improve their own models" using the distillation technique that "involves training a less capable model on the outputs of a stronger one." Anthropic said distillation can be a legitimate way to train AI systems but it's a problem when competitors "use it to acquire powerful capabilities from other labs in a fraction of the time, and at a fraction of the cost, that it would take to develop them independently." But it can go both ways. San Francisco-based startup Anysphere, maker of the popular coding tool Cursor, recently acknowledged that its latest product was based on an open-source model made by Chinese company Moonshot AI, maker of the chatbot Kimi. Kyle Chan, a fellow at the Washington-based think tank The Brookings Institution and an expert on China's technology development, said it will be like "looking for needles in an enormous haystack" to separate unauthorized distillation from the vast volume of legitimate requests for data. But information sharing and coordination among U.S. AI labs could help, and the federal government can play an important role in facilitating anti-distillation efforts across labs, Chan said. It's hard to assess how far the House bill can go, but Chan said Trump may not want to rock the boat with Chinese President Xi Jinping ahead of a planned mid-May state visit to Beijing.

The White House has Flagged Distillation Threats from Foreign AI Firms

The White House's science office said that foreign entities are using proxy accounts and jailbreaking techniques to distill capabilities from American AI models. The Trump administration has unveiled plans to combat "industrial-scale campaigns" by Chinese-based entities to copy AI technology developed by US companies. In a statement on Thursday, the assistant to the president for the White House Office of Science and Technology Policy, Michael J. Kratsios, said the government has "information" indicating that foreign entities -- primarily based in China -- are deliberately targeting major US AI firms to distill their AI models. "Models developed from surreptitious, unauthorized distillation campaigns like this do not replicate the full performance of the original. They do, however, enable foreign actors to release products that appear to perform comparably on select benchmarks at a fraction of the cost," Kratsios said. The statement from the US government comes just two months after Claude-developer Anthropic accused three Chinese AI firms of engaging in distillation attacks on its AI models. Meanwhile, the White House warned that models developed through these distillation campaigns allow actors to produce cheaper models that are stripped of security protocols, moving them away from being "neutral and truth-seeking." According to data from Morph LLM, the current per-million-token pricing of frontier AI models such as Claude's Opus 4.6 costs $5, while Chat GPT-5.4 Pro costs up to $30. China's DeepSeek V3.2, described as a mid-tier model, costs only $0.26. Tech investor Jason Calacanis said in a recent interview with the "All-In" podcast that he has been paying around $300 per day to use Anthropic AI agents to help run his businesses. The White House OSTP said the Trump administration will seek to counter threats from foreign firms by sharing information with US companies about potentially large attacks, helping the private sector better coordinate against them, working with the private sector to build strong defenses, and exploring measures to "hold foreign actors accountable." Related: Crypto-aligned Fellowship PAC bets big on Texas Senate race The White House OSTP said that the foreign companies have been using "tens of thousands of proxy accounts" to fly under the radar while "using jailbreaking techniques to expose proprietary information." "These coordinated campaigns systematically extract capabilities from American AI models, exploiting American expertise and innovation," it said. In Anthropic's case, the firm accused DeepSeek, Moonshot and MiniMax in late February of generating "over 16 million exchanges" with its AI models combined across about 24,000 "fraudulent accounts." The firm said the companies were trying to scrape Claude for capabilities such as agentic reasoning, coding and data analysis, rubric-based grading tasks, and computer vision.

Trump administration vows crackdown on Chinese companies 'exploiting' AI models made in US

WASHINGTON (AP) -- The Trump administration is vowing to crack down on foreign tech companies' exploitation of U.S. artificial intelligence models, singling out China at a time that country is narrowing the gap with the U.S. in the AI race. In a Thursday memo, Michael Kratsios, the president's chief science and technology adviser, accused foreign entities "principally based in China" of engaging in deliberate, industrial-scale campaigns to "distill," or extract capabilities from, leading AI systems made in the U.S. and "exploiting American expertise and innovation." The administration, Kratsios wrote, will work with American AI companies to identify such activities, build defenses and find ways to punish offenders. The memo arrives at a time when China is challenging U.S. dominance in artificial intelligence, an area where the White House says the U.S. must prevail to set global standards and reap economic and military benefits. But the U.S.-China gap in performance of top AI models has "effectively closed," according to a recent report from Stanford University's Institute for Human-Centered AI. China's embassy in Washington said it opposed "the unjustified suppression of Chinese companies by the U.S." "China has always been committed to promoting scientific and technological progress through cooperation and healthy competition. China attaches great importance to the protection of intellectual property rights," said Liu Pengyu, the embassy spokesperson. Kratsios' memo also came the same week that the House Foreign Affairs Committee offered unanimous, bipartisan support for a bill to set up a process to identify foreign actors that extract "key technical features" of closed-source, U.S.-owned AI models and to punish them with measures including sanctions. "Model extraction attacks are the latest frontier of Chinese economic coercion and theft of U.S. intellectual property," said Rep. Bill Huizenga, R-Mich., who sponsored the bill. "American AI models are demonstrating transformative cyber capabilities, and it is critical we prevent China from stealing these technological advancements." Last year, the Chinese start-up DeepSeek rattled U.S. markets when it released a large language model that could compete with U.S. AI giants but at a fraction of the cost. David Sacks, then serving as President Donald Trump's AI and crypto adviser, suggested that DeepSeek copied U.S. models. "There's substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI's models," Sacks said then. In a February letter to U.S. lawmakers, OpenAI, the developer of ChatGPT, made similar allegations and said China should not be allowed to advance "autocratic AI" by "appropriating and repackaging American innovation." Anthropic, the maker of the Claude chatbot, in February accused DeepSeek and two other China-based AI laboratories of engaging in campaigns to "illicitly extract Claude's capabilities to improve their own models" using the distillation technique that "involves training a less capable model on the outputs of a stronger one." Anthropic said distillation can be a legitimate way to train AI systems but it's a problem when competitors "use it to acquire powerful capabilities from other labs in a fraction of the time, and at a fraction of the cost, that it would take to develop them independently." But it can go both ways. San Francisco-based startup Anysphere, maker of the popular coding tool Cursor, recently acknowledged that its latest product was based on an open-source model made by Chinese company Moonshot AI, maker of the chatbot Kimi. Kyle Chan, a fellow at the Washington-based think tank The Brookings Institution and an expert on China's technology development, said it will be like "looking for needles in an enormous haystack" to separate unauthorized distillation from the vast volume of legitimate requests for data. But information sharing and coordination among U.S. AI labs could help, and the federal government can play an important role in facilitating anti-distillation efforts across labs, Chan said. It's hard to assess how far the House bill can go, but Chan said Trump may not want to rock the boat with Chinese President Xi Jinping ahead of a planned mid-May state visit to Beijing. ___ AP Technology Writer Matt O'Brien contributed to this report from Providence, Rhode Island.

US accuses China of 'industrial-scale' AI model theft

Washington | The Trump administration unveiled measures aimed at preventing Chinese developers from improperly using leading American AI models to build a rival generation of chatbots, marking the first major US response to Silicon Valley companies' complaints that China is piggybacking on their success. In a memo, the White House Office of Science and Technology Policy said it will promote wider information sharing by US-based developers and increase efforts to help the industry detect unauthorised extraction of their artificial intelligence models.

Trump Administration Vows Crackdown on Chinese Companies 'Exploiting' AI Models Made in US

WASHINGTON (AP) -- The Trump administration is vowing to crack down on foreign tech companies' exploitation of U.S. artificial intelligence models, singling out China at a time that country is narrowing the gap with the U.S. in the AI race. In a Thursday memo, Michael Kratsios, the president's chief science and technology adviser, accused foreign entities "principally based in China" of engaging in deliberate, industrial-scale campaigns to "distill," or extract capabilities from, leading AI systems made in the U.S. and "exploiting American expertise and innovation." The administration, Kratsios wrote, will work with American AI companies to identify such activities, build defenses and find ways to punish offenders. The memo arrives at a time when China is challenging U.S. dominance in artificial intelligence, an area where the White House says the U.S. must prevail to set global standards and reap economic and military benefits. But the U.S.-China gap in performance of top AI models has "effectively closed," according to a recent report from Stanford University's Institute for Human-Centered AI. China's embassy in Washington said it opposed "the unjustified suppression of Chinese companies by the U.S." "China has always been committed to promoting scientific and technological progress through cooperation and healthy competition. China attaches great importance to the protection of intellectual property rights," said Liu Pengyu, the embassy spokesperson. Kratsios' memo also came the same week that the House Foreign Affairs Committee offered unanimous, bipartisan support for a bill to set up a process to identify foreign actors that extract "key technical features" of closed-source, U.S.-owned AI models and to punish them with measures including sanctions. "Model extraction attacks are the latest frontier of Chinese economic coercion and theft of U.S. intellectual property," said Rep. Bill Huizenga, R-Mich., who sponsored the bill. "American AI models are demonstrating transformative cyber capabilities, and it is critical we prevent China from stealing these technological advancements." Last year, the Chinese start-up DeepSeek rattled U.S. markets when it released a large language model that could compete with U.S. AI giants but at a fraction of the cost. David Sacks, then serving as President Donald Trump's AI and crypto adviser, suggested that DeepSeek copied U.S. models. "There's substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI's models," Sacks said then. In a February letter to U.S. lawmakers, OpenAI, the developer of ChatGPT, made similar allegations and said China should not be allowed to advance "autocratic AI" by "appropriating and repackaging American innovation." Anthropic, the maker of the Claude chatbot, in February accused DeepSeek and two other China-based AI laboratories of engaging in campaigns to "illicitly extract Claude's capabilities to improve their own models" using the distillation technique that "involves training a less capable model on the outputs of a stronger one." Anthropic said distillation can be a legitimate way to train AI systems but it's a problem when competitors "use it to acquire powerful capabilities from other labs in a fraction of the time, and at a fraction of the cost, that it would take to develop them independently." But it can go both ways. San Francisco-based startup Anysphere, maker of the popular coding tool Cursor, recently acknowledged that its latest product was based on an open-source model made by Chinese company Moonshot AI, maker of the chatbot Kimi. Kyle Chan, a fellow at the Washington-based think tank The Brookings Institution and an expert on China's technology development, said it will be like "looking for needles in an enormous haystack" to separate unauthorized distillation from the vast volume of legitimate requests for data. But information sharing and coordination among U.S. AI labs could help, and the federal government can play an important role in facilitating anti-distillation efforts across labs, Chan said. It's hard to assess how far the House bill can go, but Chan said Trump may not want to rock the boat with Chinese President Xi Jinping ahead of a planned mid-May state visit to Beijing. ___ AP Technology Writer Matt O'Brien contributed to this report from Providence, Rhode Island.

White House official accuses foreign entities of 'industrial-scale' theft of US AI

The White House on Thursday accused foreign entities, primarily China, of running "industrial-scale" campaigns to distill frontier artificial intelligence models from US companies. Michael Kratsios, director of the White House Office of Science and Technology, wrote in a memo Thursday that the U.S. government has learned Chinese entities are leveraging thousands of proxy accounts and jailbreaking tactics to access proprietary information. Distillation occurs when the outputs of a stronger model are used to train less capable models. "These coordinate campaigns systematically extract capabilities from American AI models, exploiting American expertise and innovation," Kratsios wrote. The White House leader noted the models produced through this distillation do not meet the full performance capabilities of the original but allow foreign actors to release products that appear comparable for a cheaper price. These products often present national security risks as actors strip the security protocols usually embedded in U.S. versions. In response to this threat, Kratsios said the Trump administration is sharing information with US-based AI companies about the campaigns and will work with the private sector on ways to defend against such attempts. "There is nothing innovative about systematically extracting and copying the innovations of American industry, and there is nothing open about supposedly open models that are derived from acts of malicious exploitation," Kratsios wrote. The memo was first reported by The Financial Times. Tensions over technology have long simmered between the U.S. and China. The memo comes just weeks ahead of President Trump's scheduled trip to Beijing to meet with Chinese President Xi Jinping. The U.S. has sought to limit the transfer of US technology to China amid the tight AI race. Trump came under backlash from his own party late last year over his decision to allow Nvidia to sell its H200 chips to China. The H200 chips are more powerful than its H20 chips, which were developed with U.S. export controls in mind. After initially restricting H20 sales earlier this year, the Trump administration reversed course and allowed Nvidia to sell the chips to China for a 15 percent cut of revenue. The Hill reached out to the Chinese embassy for comment.

US State Dept orders global warning about alleged AI thefts by DeepSeek, other Chinese firms - The Economic Times

The US State Department has ordered a global push to bring attention to what it says are widespread efforts by Chinese companies, including AI startup DeepSeek, to steal intellectual property from U.S. artificial intelligence labs, according to a diplomatic cable seen by Reuters. The cable, dated Friday and sent to diplomatic and consular posts around the world, instructs diplomatic staff to speak to their foreign counterparts about "concerns over adversaries' extraction and distillation of U.S. A.I. models." "A separate demarche request and message has been sent to Beijing for raising with China," the document ⁠states. Distillation is ⁠the process of training smaller AI models using output from larger, more expensive ones as part of an effort to lower the costs of training a powerful new AI tool. This week, the White House made similar accusations, but the cable has not been previously reported. The State Department did not immediately respond to a request for comment. OpenAI has warned U.S. lawmakers that DeepSeek was targeting the ChatGPT maker and the nation's leading AI companies to replicate models and use them for its own training, Reuters reported in February. China rejects accusations The Chinese Embassy in Washington on Friday reiterated its stance that the accusations are baseless. "The ⁠allegations that Chinese entities are stealing American AI intellectual property are groundless and are deliberate attacks on China's development and progress in the AI industry," it said in a statement to Reuters. DeepSeek, whose low-cost AI model stunned ⁠the world last year, on Friday launched a preview of a highly anticipated new model, called the V4, adapted for Huawei chip technology, underlining China's growing autonomy in the sector. DeepSeek also did not immediately respond to a request for comment. In the past, it has said that its V3 model used naturally occurring data collected through web crawling, and it had not intentionally used synthetic data generated by OpenAI. Many Western and some Asian governments have banned their institutions and officials from using DeepSeek, citing data privacy concerns. Nevertheless, DeepSeek's models have consistently been among the most used on international platforms that host open-source models. The State Department cable said its purpose was to "warn of the risks of utilizing AI models distilled from U.S. proprietary AI models, and lay the groundwork for potential follow-up and outreach by the U.S. government." It also mentioned Chinese AI firms Moonshot AI and MiniMax. Neither company immediately responded ⁠to a request for comment. The cable said that "AI models developed from surreptitious, unauthorised distillation campaigns enable foreign actors to release products that appear to perform comparably on select benchmarks at a fraction of the cost but do not replicate the full performance of the original system." It added that the campaigns also "deliberately strip security protocols from the resulting models and undo mechanisms that ensure those AI models are ideologically neutral and truth-seeking." The White House accusations and the cable come just weeks before U.S. President Donald Trump is set to visit Chinese President Xi Jinping in Beijing. They could well raise tensions in a long-running tech war between the rival superpowers, which had been lowered by a detente brokered last October.

Trump Admin Accuses Beijing Of Industrial-Scale AI Theft From US Labs, Warns Of Crackdown On 'Distillatio

US Alleges Coordinated AI Model Copying Effort In a memo, Michael Kratsios, director of the White House Office of Science and Technology Policy, said the administration has intelligence indicating that foreign actors, primarily based in China, are running "industrial-scale campaigns" aimed at replicating advanced U.S. AI systems. Kratsios said these efforts involve extracting knowledge from leading American AI models and using it to build competing systems at lower cost, calling the activity an "unacceptable" threat when conducted without authorization. "The United States government has information indicating that foreign entities principally based in China, are engaged in deliberate, industrial-scale campaigns to distil US frontier AI systems," he wrote. Distillation At Center Of AI Security Concerns The practice, known as distillation, is a standard AI technique used to create smaller, efficient models. U.S. AI companies, including Anthropic and OpenAI, have previously raised growing concerns that distillation efforts by Chinese groups could erode America's technological lead. Kratsios said such campaigns often rely on "tens of thousands of proxy accounts" and other methods to evade detection while attempting to extract proprietary model behavior. National Security And Industry Response Chris McGuire of the Council on Foreign Relations told the Financial Times that Chinese firms are using distillation to "offset deficits in AI computing power and illicitly reproduce the core capabilities of U.S. models." Crackdown And Policy Measures The administration said it will share intelligence with AI companies, strengthen coordination, and explore sanctions or export restrictions against entities involved. Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors. Photo courtesy: Shutterstock Market News and Data brought to you by Benzinga APIs To add Benzinga News as your preferred source on Google, click here.

Trump administration vows crackdown on Chinese companies 'exploiting' AI models made in US - The Economic Times

America is taking a strong stance against foreign tech firms, especially those in China, for allegedly stealing advanced artificial intelligence capabilities. The US government plans to work with domestic AI companies to identify and counter these activities.The Trump administration is vowing to crack down on foreign tech companies' exploitation of US artificial intelligence models, singling out China at a time that country is narrowing the gap with the US in the AI race. In a Thursday memo, Michael Kratsios, the president's chief science and technology adviser, accused foreign entities "principally based in China" of engaging in deliberate, industrial-scale campaigns to "distill," or extract capabilities from, leading AI systems made in the U.S. and "exploiting American expertise and innovation." The administration, Kratsios wrote, will work with American AI companies to identify such activities, build defenses and find ways to punish offenders. The memo arrives at a time when China is challenging U.S. dominance in artificial intelligence, an area where the White House says the U.S. must prevail to set global standards and reap economic and military benefits. But the U.S.-China gap in performance of top AI models has "effectively closed," according to a recent report from Stanford University's Institute for Human-Centered AI. China's embassy in Washington said it opposed "the unjustified suppression of Chinese companies by the US." "China has always been committed to promoting scientific and technological progress through cooperation and healthy competition. China attaches great importance to the protection of intellectual property rights," said Liu Pengyu, the embassy spokesperson. In Beijing, China's Foreign Ministry spokesperson Guo Jiakun told reporters Friday that the U.S. claims are groundless and were smearing the achievements of China's artificial intelligence industry. "China firmly opposes this. We urge the US to respect facts, discard prejudice, stop suppressing China's technological development, and do more to promote scientific and technological exchange and cooperation between the two countries," he said. Kratsios' memo also came the same week that the House Foreign Affairs Committee offered unanimous, bipartisan support for a bill to set up a process to identify foreign actors that extract "key technical features" of closed-source, U.S.-owned AI models and to punish them with measures including sanctions. "Model extraction attacks are the latest frontier of Chinese economic coercion and theft of U.S. intellectual property," said Rep. Bill Huizenga, R-Mich., who sponsored the bill. "American AI models are demonstrating transformative cyber capabilities, and it is critical we prevent China from stealing these technological advancements." Last year, the Chinese startup DeepSeek rattled U.S. markets when it released a large language model that could compete with U.S. AI giants but at a fraction of the cost. David Sacks, then serving as President Donald Trump's AI and crypto adviser, suggested that DeepSeek copied U.S. models. "There's substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI's models," Sacks said then. In a February letter to U.S. lawmakers, OpenAI, the developer of ChatGPT, made similar allegations and said China should not be allowed to advance "autocratic AI" by "appropriating and repackaging American innovation." Anthropic, the maker of the Claude chatbot, in February accused DeepSeek and two other China-based AI laboratories of engaging in campaigns to "illicitly extract Claude's capabilities to improve their own models" using the distillation technique that "involves training a less capable model on the outputs of a stronger one." Anthropic said distillation can be a legitimate way to train AI systems but it's a problem when competitors "use it to acquire powerful capabilities from other labs in a fraction of the time, and at a fraction of the cost, that it would take to develop them independently." But it can go both ways. San Francisco-based startup Anysphere, maker of the popular coding tool Cursor, recently acknowledged that its latest product was based on an open-source model made by Chinese company Moonshot AI, maker of the chatbot Kimi. Kyle Chan, a fellow at the Washington-based think tank The Brookings Institution and an expert on China's technology development, said it will be like "looking for needles in an enormous haystack" to separate unauthorized distillation from the vast volume of legitimate requests for data. But information sharing and coordination among U.S. AI labs could help, and the federal government can play an important role in facilitating anti-distillation efforts across labs, Chan said. It's hard to assess how far the House bill can go, but Chan said Trump may not want to rock the boat with Chinese President Xi Jinping ahead of a planned mid-May state visit to Beijing.

White House Accuses China of Far-Reaching Theft of AI Tech | PYMNTS.com

With that accusation comes the warning that the government will crack down on this practice, the Financial Times (FT) reported Thursday (April 23), citing a memo seen by the news outlet. "The US government has information indicating that foreign entities, principally based in China, are engaged in deliberate, industrial-scale campaigns to distill US frontier AI systems," wrote Michael Kratsios, director of the White House Office of Science and Technology Policy. As the FT noted, the accusation marks an escalation in tensions around Chinese companies allegedly using American AI research in the race to dominate the technology, and is happening ahead of a meeting between the countries' presidents. The issue came to attention when China's DeepSeek was accused of using distillation -- or training smaller AI models based on the output of larger ones -- to develop a powerful product at a lower cost. Kratsios' memo said the Trump administration would share information with American AI companies about "attempts by foreign actors to conduct unauthorized, industrial-scale distillation" and help them prevent attacks. He added that Chinese campaigns were "leveraging tens of thousands of proxy accounts to evade detection and using jailbreaking techniques to expose proprietary information." The White House's warning follows reports last week about proposed legislation that would crack down on Chinese companies that copy American-made AI models. Rep. Bill Huizenga's (R-MI) bill would sanction entities in China and Russia engaged in using improper "query-and-copy" techniques on American AI models. "Model extraction attacks are the latest frontier of Chinese economic coercion and theft of US intellectual property," Huizenga, who sits on the House Foreign Affairs Committee, said in a statement included in a Bloomberg News report on the issue. Several Chinese AI firms -- DeepSeek included -- have been accused by American AI startup Anthropic of illicitly using the company's output to train their models. The company said in February it had found a total of 24,000 fraudulent accounts that the three labs used to generate 16 million exchanges with its Claude model, violating Anthropic's terms of service and regional access restrictions. "These campaigns are growing in intensity and sophistication," Anthropic said in a blog post. "The window to act is narrow, and the threat extends beyond any single company or region. Addressing it will require rapid, coordinated action among industry players, policymakers and the global AI community."

Trump administration vows crackdown on Chinese companies 'exploiting' AI models made in US

WASHINGTON: The Trump administration is vowing to crack down on foreign tech companies' exploitation of U.S. artificial intelligence models, singling out China at a time that country is narrowing the gap with the U.S. in the AI race. In a Thursday memo, Michael Kratsios, the president's chief science and technology adviser, accused foreign entities "principally based in China" of engaging in deliberate, industrial-scale campaigns to "distill," or extract capabilities from, leading AI systems made in the U.S. and "exploiting American expertise and innovation." Also Read: Donald Trump's DOJ has cut thousands of law-enforcement jobs while vowing to get tough on crime The administration, Kratsios wrote, will work with American AI companies to identify such activities, build defenses and find ways to punish offenders. The memo arrives at a time when China is challenging U.S. dominance in artificial intelligence, an area where the White House says the U.S. must prevail to set global standards and reap economic and military benefits. But the U.S.-China gap in performance of top AI models has "effectively closed," according to a recent report from Stanford University's Institute for Human-Centered AI. China's embassy in Washington said it opposed "the unjustified suppression of Chinese companies by the U.S." "China has always been committed to promoting scientific and technological progress through cooperation and healthy competition. China attaches great importance to the protection of intellectual property rights," said Liu Pengyu, the embassy spokesperson. Kratsios' memo also came the same week that the House Foreign Affairs Committee offered unanimous, bipartisan support for a bill to set up a process to identify foreign actors that extract "key technical features" of closed-source, U.S.-owned AI models and to punish them with measures including sanctions. "Model extraction attacks are the latest frontier of Chinese economic coercion and theft of U.S. intellectual property," said Rep. Bill Huizenga, R-Mich., who sponsored the bill. "American AI models are demonstrating transformative cyber capabilities, and it is critical we prevent China from stealing these technological advancements." Also Read: US invites Putin to G20 summit but Trump doubts he'll come Last year, the Chinese start-up DeepSeek rattled U.S. markets when it released a large language model that could compete with U.S. AI giants but at a fraction of the cost. David Sacks, then serving as President Donald Trump's AI and crypto adviser, suggested that DeepSeek copied U.S. models. "There's substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI's models," Sacks said then. In a February letter to U.S. lawmakers, OpenAI, the developer of ChatGPT, made similar allegations and said China should not be allowed to advance "autocratic AI" by "appropriating and repackaging American innovation." Anthropic, the maker of the Claude chatbot, in February accused DeepSeek and two other China-based AI laboratories of engaging in campaigns to "illicitly extract Claude's capabilities to improve their own models" using the distillation technique that "involves training a less capable model on the outputs of a stronger one." Anthropic said distillation can be a legitimate way to train AI systems but it's a problem when competitors "use it to acquire powerful capabilities from other labs in a fraction of the time, and at a fraction of the cost, that it would take to develop them independently." But it can go both ways. San Francisco-based startup Anysphere, maker of the popular coding tool Cursor, recently acknowledged that its latest product was based on an open-source model made by Chinese company Moonshot AI, maker of the chatbot Kimi. Kyle Chan, a fellow at the Washington-based think tank The Brookings Institution and an expert on China's technology development, said it will be like "looking for needles in an enormous haystack" to separate unauthorized distillation from the vast volume of legitimate requests for data. But information sharing and coordination among U.S. AI labs could help, and the federal government can play an important role in facilitating anti-distillation efforts across labs, Chan said. It's hard to assess how far the House bill can go, but Chan said Trump may not want to rock the boat with Chinese President Xi Jinping ahead of a planned mid-May state visit to Beijing.

Trump administration vows crackdown on Chinese companies 'exploiting' AI models made in U.S.

WASHINGTON -- The Trump administration is vowing to crack down on foreign tech companies' exploitation of U.S. artificial intelligence models, singling out China at a time that country is narrowing the gap with the U.S. in the AI race. In a Thursday memo, Michael Kratsios, the president's chief science and technology adviser, accused foreign entities "principally based in China" of engaging in deliberate, industrial-scale campaigns to "distill," or extract capabilities from, leading AI systems made in the U.S. and "exploiting American expertise and innovation." The administration, Kratsios wrote, will work with American AI companies to identify such activities, build defenses and find ways to punish offenders. The memo arrives at a time when China is challenging U.S. dominance in artificial intelligence, an area where the White House says the U.S. must prevail to set global standards and reap economic and military benefits. But the U.S.-China gap in performance of top AI models has "effectively closed," according to a recent report from Stanford University's Institute for Human-Centered AI. China's embassy in Washington said it opposed "the unjustified suppression of Chinese companies by the U.S." "China has always been committed to promoting scientific and technological progress through cooperation and healthy competition. China attaches great importance to the protection of intellectual property rights," said Liu Pengyu, the embassy spokesperson. In Beijing, China's Foreign Ministry spokesperson Guo Jiakun told reporters Friday that the U.S. claims are groundless and were smearing the achievements of China's artificial intelligence industry. "China firmly opposes this. We urge the U.S. to respect facts, discard prejudice, stop suppressing China's technological development, and do more to promote scientific and technological exchange and cooperation between the two countries," he said. Kratsios' memo also came the same week that the House Foreign Affairs Committee offered unanimous, bipartisan support for a bill to set up a process to identify foreign actors that extract "key technical features" of closed-source, U.S.-owned AI models and to punish them with measures including sanctions. "Model extraction attacks are the latest frontier of Chinese economic coercion and theft of U.S. intellectual property," said Rep. Bill Huizenga, R-Mich., who sponsored the bill. "American AI models are demonstrating transformative cyber capabilities, and it is critical we prevent China from stealing these technological advancements." Last year, the Chinese startup DeepSeek rattled U.S. markets when it released a large language model that could compete with U.S. AI giants but at a fraction of the cost. David Sacks, then serving as U.S. President Donald Trump's AI and crypto adviser, suggested that DeepSeek copied U.S. models. "There's substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI's models," Sacks said then. In a February letter to U.S. lawmakers, OpenAI, the developer of ChatGPT, made similar allegations and said China should not be allowed to advance "autocratic AI" by "appropriating and repackaging American innovation." Anthropic, the maker of the Claude chatbot, in February accused DeepSeek and two other China-based AI laboratories of engaging in campaigns to "illicitly extract Claude's capabilities to improve their own models" using the distillation technique that "involves training a less capable model on the outputs of a stronger one." Anthropic said distillation can be a legitimate way to train AI systems but it's a problem when competitors "use it to acquire powerful capabilities from other labs in a fraction of the time, and at a fraction of the cost, that it would take to develop them independently." But it can go both ways. San Francisco-based startup Anysphere, maker of the popular coding tool Cursor, recently acknowledged that its latest product was based on an open-source model made by Chinese company Moonshot AI, maker of the chatbot Kimi. Kyle Chan, a fellow at the Washington-based think tank The Brookings Institution and an expert on China's technology development, said it will be like "looking for needles in an enormous haystack" to separate unauthorized distillation from the vast volume of legitimate requests for data. But information sharing and coordination among U.S. AI labs could help, and the federal government can play an important role in facilitating anti-distillation efforts across labs, Chan said. It's hard to assess how far the House bill can go, but Chan said Trump may not want to rock the boat with Chinese President Xi Jinping ahead of a planned mid-May state visit to Beijing.

China stealing U.S. AI technology: White House official

The White House on Thursday accused Chinese entities of a massive effort to steal U.S. artificial intelligence technology and vowed to take action to prevent the alleged theft. "The U.S. has evidence that foreign entities, primarily in China, are running industrial-scale distillation campaigns to steal American AI," White House science and technology chief Michael Kratsios said in a post on X. "We will be taking action to protect American innovation." Distillation is a common practice within AI development, often used by companies to create cheaper, smaller versions of their own models. In February, U.S. AI developer Anthropic accused three Chinese firms, DeepSeek, Moonshot AI and MiniMax, of running campaigns to illicitly extract capabilities from its Claude chatbot, describing it as intellectual property theft. That same month, ChatGPT creator OpenAI sent a letter to U.S. legislators accusing DeepSeek of using distillation techniques amid "ongoing efforts to free-ride on the capabilities developed by OpenAI and other U.S. frontier labs." Kratsios did not name any specific foreign entities in his post but said they "are using tens of thousands of proxies and jailbreaking techniques in coordinated campaigns to systematically extract American breakthroughs." The accusations come ahead of a planned May 14 summit in Beijing between U.S. President Donald Trump and his Chinese counterpart Xi Jinping.

US State Dept orders global warning about alleged AI thefts by DeepSeek, other Chinese firms

WASHINGTON, April 24 (Reuters) - The U.S. State Department has ordered a global push to bring attention to what it says are widespread efforts by Chinese companies, including AI startup DeepSeek, to steal intellectual property from U.S. artificial intelligence labs, according to a diplomatic cable seen by Reuters. The cable, dated Friday and sent to diplomatic and consular posts around the world, instructs diplomatic staff to speak to their foreign counterparts about "concerns over adversaries' extraction and distillation of U.S. A.I. models." "A separate demarche request and message has been sent to Beijing for raising with China," the document states. Distillation is the process of training smaller AI models using output from larger, more expensive ones as part of an effort to lower the costs of training a powerful new AI tool. This week, the White House made similar accusations, but the cable has not been previously reported. The State Department did not immediately respond to a request for comment. OpenAI has warned U.S. lawmakers that DeepSeek was targeting the ChatGPT maker and the nation's leading AI companies to replicate models and use them for its own training, Reuters reported in February. CHINA REJECTS ACCUSATIONS The Chinese Embassy in Washington on Friday reiterated its stance that the accusations are baseless. "The allegations that Chinese entities are stealing American AI intellectual property are groundless and are deliberate attacks on China's development and progress in the AI industry," it said in a statement to Reuters. DeepSeek, whose low-cost AI model stunned the world last year, on Friday launched a preview of a highly anticipated new model, called the V4, adapted for Huawei chip technology, underlining China's growing autonomy in the sector. DeepSeek also did not immediately respond to a request for comment. In the past, it has said that its V3 model used data naturally occurring and collected through web crawling and it had not intentionally used synthetic data generated by OpenAI. Many Western and some Asian governments have banned their institutions and officials from using DeepSeek, citing data privacy concerns. Nevertheless, DeepSeek's models have consistently been among the most used on international platforms that host open-source models. The State Department cable said its purpose was to "warn of the risks of utilizing AI models distilled from U.S. proprietary AI models, and lay the groundwork for potential follow-up and outreach by the U.S. government." It also mentioned Chinese AI firms Moonshot AI and MiniMax. Neither company immediately responded to a request for comment. The cable said that "AI models developed from surreptitious, unauthorized distillation campaigns enable foreign actors to release products that appear to perform comparably on select benchmarks at a fraction of the cost but do not replicate the full performance of the original system." It added that the campaigns also "deliberately strip security protocols from the resulting models and undo mechanisms that ensure those AI models are ideologically neutral and truth-seeking." The White House accusations and the cable come just ?weeks before U.S. President Donald Trump is set to visit Chinese President Xi Jinping in Beijing. They could well raise tensions in a long-running tech war between the rival superpowers, which had been lowered by a detente brokered last October. (Reporting by Raphael Satter and Alexandra Alper in Washington; Editing by Chris Sanders, David Gregorio and Edwina Gibbs)

US State Dept orders global warning about alleged China AI thefts by DeepSeek, others

WASHINGTON, April 24 (Reuters) - The U.S. State Department has ordered a global push to bring attention to what it says are widespread efforts by Chinese companies, including AI startup DeepSeek, to steal intellectual property from U.S. artificial intelligence labs, according to a diplomatic cable seen by Reuters. The cable said its purpose was to "warn of the risks of utilizing AI models distilled from U.S. proprietary AI models, and lay the groundwork for potential follow-up and outreach by the U.S. government." Distillation is the process of training smaller AI models using output from larger, more expensive ones as part of an effort to lower the costs of training a powerful new AI tool. DeepSeek, the Chinese startup whose low-cost AI model stunned the world last year, on Friday launched a preview of a highly anticipated new model adapted for Huawei chip technology, underlining China's growing autonomy in the sector. The State Department, DeepSeek and the Chinese Embassy in Washington did not immediately respond to a request for comment. The cable also mentioned Chinese AI firms Moonshot AI and MiniMax. Neither company immediately responded to a request for comment. This week, the White House made similar accusations, which the Chinese Embassy in Washington called "baseless allegations," adding that Beijing "attaches great importance to the protection of intellectual property rights." The cable, dated Friday and sent to diplomatic and consular posts around the world, instructs diplomatic staff to speak to their foreign counterparts about "concerns over adversaries' extraction and distillation of U.S. A.I. models." "A separate demarche request and message has been sent to Beijing for raising with China," the document states. The cable, which has not been previously reported, signals the Trump administration is taking seriously the growing concerns about Chinese distillation of U.S. AI models. "AI models developed from surreptitious, unauthorized distillation campaigns enable foreign actors to release products that appear to perform comparably on select benchmarks at a fraction of the cost but do not replicate the full performance of the original system," the cable said, adding that the campaigns also "deliberately strip security protocols from the resulting models and undo mechanisms that ensure those AI models are ideologically neutral and truth-seeking." OpenAI warned U.S. lawmakers that Chinese artificial intelligence startup DeepSeek was targeting the ChatGPT maker and the nation's leading AI companies to replicate models and use them for its own training, Reuters reported in February. The memo and follow-up cable, released just ?weeks before U.S. President Donald Trump is set to visit Chinese President Xi Jinping in Beijing, promise to raise tensions in a long-running tech war between the rival superpowers, which had been lowered by a detente brokered last October. (Reporting by Raphael Satter and Alexandra Alper in Washington; editing by Chris Sanders and David Gregorio)