Trump will never back a US AI regulator, says departing tech adviser Sriram Krishnan

Trump will never back a US AI regulator, says outgoing tech adviser

Donald Trump will not establish a formal licensing regime for AI, the president's departing AI adviser has said, even as the White House wields emergency powers to stall the most advanced models. "There will not be an FDA for AI," Sriram Krishnan told the FT in his first in-depth interview since leaving the administration last month, referring to the powerful US food and drug regulator. "This administration, [the] president, from day one has been against burdensome, onerous, bureaucratic red tape," he added. "We are not in the business of picking winners and losers." Krishnan said setting up a centralised agency requiring "a team of lawyers before you can get a model out" would put "sand in the gears" of the AI revolution. "That is never, never going to happen under President Trump," he said. His prediction comes weeks after the US government made an unprecedented intervention on national security grounds to force Anthropic to withdraw its most capable model, Mythos, and hold up the release of OpenAI's 5.6. Krishnan, an Indian-born former venture capitalist who worked with Elon Musk before entering government, has been among the key advocates for light-touch AI regulation in Trump's circle alongside AI tsar David Sacks. Some in Washington blame the administration's deregulatory zeal for a growing backlash against AI among American voters and within Trump's Maga coalition, reflecting concern about data centres and the societal impact of the technology. A large majority of Americans support stringent regulations to rein in AI, while at least 75 data centre projects worth approximately $130bn were disrupted by local opposition in the first three months of 2026, according to researcher Data Center Watch. But Krishnan, who has worked to block state-level AI regulation and helped draft a White House framework for narrow child safety and content laws, laid the blame for the backlash squarely on the industry's own "doomer" messaging. The AI sector "has done a terrible job" of explaining the benefits of the tech, Krishnan said, such as advanced medical diagnoses. Leaders of American AI labs "have focused so much on the dystopian narrative and scenarios, whether it is job loss, whether it is [existential] risk . . . that a lot of people are going: well, I don't know if I want this," he added. Voters, he argued, wanted to feel "they have a seat at the table and this is not just two or three companies becoming incredibly powerful and wealthy" -- a sentiment the president has seized on. Trump recently pushed to get companies to donate equity stakes to the American people, and discussed the plans with OpenAI chief executive Sam Altman. Despite warnings from Silicon Valley about this amounting to backdoor nationalisation, Krishnan supported the move. "There are lots of questions about what form it can take . . . but I do think some way for regular people to feel like, OK, when I use this model, or when I see the graphs going up and I watch on CNBC, I am benefiting. I think that is good." Americans, he added, "need to feel that AI is something which is empowering them". The 42-year-old, who entered the White House days after Trump's inauguration in 2025, was one of a small group of Silicon Valley figures who moved to Washington early in the president's second term to help the administration develop its approach to AI. A former Andreessen Horowitz partner, he spent his first 24 hours in office briefing the president and the National Security Council on the capabilities of China's DeepSeek models and writing the phrase "we need to unleash American entrepreneurship" into a Trump speech that evening. Among his strange duties as Silicon Valley's translator in Washington, he said, was explaining Roko's Basilisk, an internet thought experiment about a vengeful future superintelligence, "to a bunch of the cabinet". Over the past 18 months, Krishnan and Sacks pushed for a ban on "onerous" state-level AI safeguards, and a light-touch approach to monitoring advanced models for national and cyber security risks. The duo, who are close to Trump mega-donor Marc Andreessen, warned that throttling AI innovation with excessive regulation would help Chinese companies eclipse America's leading models. But more recently a series of new models with powerful cyber security capabilities have driven renewed momentum for the government to take a hand in controlling the technology. OpenAI's Altman last week said the government's demand to limit the release of its latest 5.6 model was not "optimal", while Silicon Valley investors warned against the creation of an ad hoc licensing regime in which frontier models must get clearance from the Trump administration. Some accused the government of seeking to punish Anthropic, which Trump has called "radical left" over its clash with the Pentagon about the use of its AI in warfare -- an allegation that Krishnan denied. Krishnan has spent the past few weeks speaking to world leaders unnerved by Washington's unilateral move to switch off access to Mythos. He supported Trump's use of export controls to force Anthropic to temporarily suspend the model, but said the government took the decision "very, very reluctantly", after safety concerns were raised by Amazon, which had found a vulnerability in the model. "I think the government will always try to make sure our systems and our security [are] protected," Krishnan said, adding that senior administration figures like vice-president JD Vance raised concerns about shielding critical infrastructure from advanced models. In the longer term, Krishnan argued, oversight of frontier models should pass largely to the industry itself, with a voluntary body or "clearinghouse" that includes large tech groups, chipmakers, and security companies that would seek to identify vulnerabilities "in concert with the intelligence community or the Department of War". Such a system would comply with a recent executive order from the president, which established a voluntary framework through which the US government would get 30 days to review a model before its release. But Krishnan warned that if cutting-edge AI tools were held back by the government for several weeks, "that would probably be bad for American innovation". The departing Trump adviser leaves the administration months after Sacks ended his role as AI tsar and is now recruiting AI engineers to help governments across the west develop AI policy. Krishnan added that US allies might seek to use "open weight" models with public training parameters that they can fine-tune. "It is very concerning to me that we have these Chinese open-weight models that are good, and we don't really have a leading American open-weight model yet," he said. Asked whether a future Democratic government could use the Trump administration's unilateral use of export controls as a pretext to stall the rollout of AI, Krishnan said: "I don't think about future governments. I think about this government and this moment in time."

Trump Won't Create AI Regulator, Says Former White House Adviser Sriram Krishnan Amid Data Center Backlas

Former Senior White House Policy Advisor on Artificial Intelligence Sriram Krishnan said the White House will not create a centralized AI licensing body akin to the Food and Drug Administration. Krishnan Rules Out AI Licensing "This administration... has been against burdensome, onerous, bureaucratic red tape," Krishnan told the Financial Times, as reported on Friday. He added officials aren't "picking winners and losers." According to the report, Krishnan also said setting up a centralized agency to license AI models would put "sand in the gears" of the industry, requiring "a team of lawyers before you can get a model out." He added flatly: "That is never, never going to happen under President Trump." Krishnan Blames Industry, Not Policy, for AI Backlash Billionaire investor Mark Cuban made a similar point in June, saying data center opposition is really "a proxy for the hate towards AI and the concentration and accumulation of wealth it's creating." Industry Self-Policing Favored Over Formal Rules Krishnan pointed to a recent executive order creating a voluntary framework giving the government 30 days to review a model before release. In the longer term, he favors shifting oversight to an industry-run "clearinghouse" that works with intelligence and defense officials, rather than formal regulation. He warned that holding back AI tools for weeks "would probably be bad for American innovation," and has separately argued that excessive regulation could let Chinese companies overtake U.S. AI labs. Krishnan Defends Anthropic's Mythos Halt Photo Courtesy: Shutterstock Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors. Market News and Data brought to you by Benzinga APIs To add Benzinga News as your preferred source on Google, click here.