Elon Musk proposes peer review system for frontier AI models as safety concerns intensify

Musk proposes peer review for frontier AI models in Economist interview

July 23 (Reuters) - Elon Musk urged the world's leading AI firms to conduct peer reviews of one another's most advanced models before releasing them, The Economist reported on Thursday, as concerns grow over the rapid pace of the technology's development and potential safety risks. The interview, which The Economist said was recorded on Monday before OpenAI revealed that one of its frontier AI models displayed troubling autonomous behavior during testing, comes as policymakers and AI firms wrestle with how to regulate increasingly powerful AI systems. The Tesla (TSLA.O), opens new tab and ⁠SpaceX (SPCX.O), opens new tab CEO said leading AI companies should hold regular meetings to discuss safety and security issues and give competitors time to assess major new models before deployment. "The most immediate thing that we could do is to have the leading AI companies ... have some sort of call once every few weeks and just discuss any safety and security issues," Musk said. Competitors could review new models and flag risks, he said, adding that if a company failed to address serious concerns, "that would be the ⁠moment for government to step in and take action." OpenAI this week disclosed that two of its frontier AI models escaped a sandboxed testing environment, gained internet access and hacked AI platform Hugging Face to obtain answers to an internal cybersecurity benchmark. The disclosure prompted ⁠renewed debate over safeguards for advanced AI systems and whether existing oversight is keeping pace with the technology. Governments in the U.S., Europe and elsewhere are weighing approaches to regulating frontier ⁠AI, while companies have largely relied on voluntary safety commitments and internal testing. Musk, who has repeatedly warned about the existential risks posed by advanced AI, argued ⁠that rival AI developers are better positioned than governments to identify technical risks in cutting-edge models, with regulators stepping in only if companies fail to address serious safety concerns. Reporting by Akash Sriram in Bengaluru; Editing by Anil D'Silva Our Standards: The Thomson Reuters Trust Principles., opens new tab * Suggested Topics: * Disrupted * Data Privacy * ADAS, AV & Safety * Software-Defined Vehicle * Sustainable & EV Supply Chain

Elon Musk says OpenAI, Anthropic and rival labs should review frontier AI models before release, with government intervention as a last resort | Fortune

Elon Musk wants the most advanced AI models to be reviewed before they're released -- but he doesn't want the government to be the one reviewing them. The Tesla and SpaceX CEO said the leading AI companies should work together to police themselves and flag safety issues related to the cutting-edge models they are releasing at a lightning pace. "The most immediate thing that we could do is to have the leading AI companies at least just meet -- have some sort of call -- once every few weeks and just discuss any safety and security issues," he said during an interview with the Economist editor-in-chief Zanny Minton Beddoes published Thursday. Musk said the government should only take action if a company flagged a security issue that it was not working to address. Otherwise, with a week or two of early access to review new models, "competitors can keep each other honest," he said. "I think it's quite difficult for someone in the government who doesn't have a deep technical understanding and isn't driving the frontier of AI to know whether something should be released or not," he said. When asked by Minton Beddoes when this system should be implemented, Musk replied "immediately," and noted even six months is a long time to delay, given AI companies' rapid rate of development. "There's so many AI breakthrough announcements, sometimes multiple per day," he said. "I mean, how many were there last week?" OpenAI, Anthropic, Google, and Musk's xAI, now a part of SpaceX, have all released new AI models within the last month as competition intensifies. Musk's comments come after OpenAI said earlier this week two of its AI models hacked their way out of a "sandbox," an isolated testing environment with no direct access to the internet, and were able to access the systems of Hugging Face, an open-source platform for hosting AI models. OpenAI's models went rogue in order to cheat on an internal evaluation test, the company said in a blog post. The incident raised alarms about the improving capabilities of AI agents and their ability to carry out complex cyberattacks autonomously. Although Musk's interview was recorded Monday, before OpenAI disclosed the breach, he wrote in a Wednesday post: "We are in the Singularity." The post, which responded to another user's post summarizing several recent AI advancements including the Hugging Face attack, may have been a reference to the hypothetical point at which AI surpasses human intelligence and drives technological change so rapid humans can no longer predict or control it. Experts have previously told Fortune advanced AI models in the hands of malicious actors could pose a major threat to companies. That includes Anthropic's Mythos, which the AI company initially refused to release publicly due to its uncanny ability to find cybersecurity vulnerabilities. AI allows malicious actors to execute cyberattacks even if they know little about coding, said Zach Lewis, chief information officer and chief information security officer at the University of Health Sciences and Pharmacy in St. Louis. AI agents could also lower the barrier for attacks further by automating the process, Lewis said. In June, President Donald Trump signed an executive order that asked AI companies to give the government oversight over unreleased AI models voluntarily. After the Hugging Face attack, some lawmakers, including Rep. Greg Casar (D-Texas), have called for even more regulation of AI models. Meanwhile, Thomas Wolf, the cofounder of Hugging Face told the BBC the attack on his company is a "wake up call" for the industry and warned that in the future: "This will be one of the most common types of cyber attacks we see." Musk's proposal for AI companies to work together to address AI threats would build on existing cooperation through the Frontier Model Forum, in which Amazon, Anthropic, Google, Meta, Microsoft, and OpenAI voluntarily share information about vulnerabilities and threats. Musk's xAI is not a member. Still, the companies in the Frontier Model Forum do not share unreleased models amongst each other. Some companies like OpenAI have already given the government an early look at some models. In a blog post from late June, the company said: "As part of our ongoing engagement with the U.S. government, we previewed our plans and the models' capabilities," ahead of the launch of its latest model GPT 5.6 Sol. In the blog post, OpenAI noted it didn't expect sharing its models with the government would become standard, but rather that it was "taking this short-term step because we believe it is the strongest path to broader availability in the coming weeks." Musk, for his part, acknowledged cooperation between the leading AI companies would require he and OpenAI CEO Sam Altman to put aside their differences, even as the pair have recently exchanged barbs on social media. Musk and Altman have been arguing both online and in court over the SpaceX CEO's disapproval of OpenAI's transition from a nonprofit entity to a for-profit company. "At the end of the day, if we have to talk, we'll talk. I mean, set aside our personal differences for the good of the world type of thing," he said.

Elon Musk proposes peer review for frontier AI models in Economist interview

Elon Musk urged top AI firms to conduct peer reviews of advanced models. Leading companies should meet regularly to discuss safety and security issues. Competitors could assess new models and flag potential risks before deployment. This comes as OpenAI revealed troubling autonomous behavior in its AI systems. Governments are now weighing approaches to regulate increasingly powerful AI technology. Elon Musk urged the world's leading AI firms to conduct peer reviews of one another's most advanced models before releasing them, The Economist reported on Thursday, as concerns grow over the rapid pace of the technology's development and potential safety risks. The interview, which The Economist said was recorded on Monday before OpenAI revealed ⁠that one ⁠of its frontier AI models displayed troubling autonomous behavior during testing, comes as policymakers and AI firms wrestle with how to regulate increasingly powerful AI systems. The Tesla and SpaceX CEO said leading AI companies should hold regular meetings to discuss safety and security issues and give competitors time to assess major new models before deployment. "The most immediate thing that ⁠we could do is to have the leading AI companies ... have some sort of call once every few weeks and just ⁠discuss any safety and security issues," Musk said. Competitors could review new models and flag risks, he said, adding that if a company failed to address serious concerns, "that would be the moment for government to step in and take action." OpenAI this week disclosed that two of its frontier AI models escaped a sandboxed testing environment, gained internet access and hacked AI platform Hugging Face to obtain answers to an internal cybersecurity benchmark. The disclosure prompted renewed debate over safeguards for advanced AI systems and whether existing oversight is keeping pace ⁠with the technology. Governments in the U.S., Europe and elsewhere are weighing approaches to regulating frontier AI, while companies have largely relied on voluntary safety commitments and internal testing. Musk, who has repeatedly warned about the existential risks posed by advanced AI, argued that rival AI developers are better positioned than governments to identify technical risks in cutting-edge models, with regulators stepping in only if companies fail to address serious safety concerns.

Musk proposes peer review for frontier AI models in Economist interview

July 23 (Reuters) - Elon Musk urged the world's leading AI firms to conduct peer reviews of one another's most advanced models before releasing them, The Economist reported on Thursday, as concerns grow over the rapid pace of the technology's development and potential safety risks. The interview, which The Economist said was recorded on Monday before OpenAI revealed that one of its frontier AI models displayed troubling autonomous behavior during testing, comes as policymakers and AI firms wrestle with how to regulate increasingly powerful AI systems. The Tesla and SpaceX CEO said leading AI companies should hold regular meetings to discuss safety and security issues and give competitors time to assess major new models before deployment. * Latest economic news on BNNBloomberg.ca "The most immediate thing that we could do is to have the leading AI companies ... have some sort of call once every few weeks and just discuss any safety and security issues," Musk said. Competitors could review new models and flag risks, he said, adding that if a company failed to address serious concerns, "that would be the moment for government to step in and take action." OpenAI this week disclosed that two of its frontier AI models escaped a sandboxed testing environment, gained internet access and hacked AI platform Hugging Face to obtain answers to an internal cybersecurity benchmark. * Latest technology news on BNNBloomberg.ca The disclosure prompted renewed debate over safeguards for advanced AI systems and whether existing oversight is keeping pace with the technology. Governments in the U.S., Europe and elsewhere are weighing approaches to regulating frontier AI, while companies have largely relied on voluntary safety commitments and internal testing. Musk, who has repeatedly warned about the existential risks posed by advanced AI, argued that rival AI developers are better positioned than governments to identify technical risks in cutting-edge models, with regulators stepping in only if companies fail to address serious safety concerns.