Microsoft, Google and xAI will let the government test their AI models before launch

In a significant move aimed at bolstering cybersecurity, tech leaders Google, Microsoft, and xAI have agreed to share their unreleased AI models with the U.S. government. This initiative, unveiled by the National Institute of Standards and Technology on Tuesday, is designed to address rising concerns about the potential risks posed by advanced AI technologies. The partnership comes in the wake of fears triggered by the recent release of Anthropic’s Mythos AI model, which has been deemed a game-changer in its capabilities. This has led the White House to consider implementing a formal review process for AI systems before they become publicly available. Under the new agreements, the Center for AI Standards and Innovation (CAISI), part of the Department of Commerce, will assess these models for their implications on national security and public safety prior to their launch. CAISI has already conducted over 40 evaluations of AI models, focusing on their real-world impacts. Chris Fall, the director of CAISI, emphasized the importance of rigorous scientific measurement in understanding the implications of frontier AI technologies. "These expanded industry collaborations allow us to enhance our efforts in the public interest at a pivotal time," he stated. The concerns surrounding Mythos have led various sectors, including government, finance, and utilities, to express apprehension about AI's cybersecurity implications. Anthropic has opted to restrict access to its model, sharing it only with a select group of approved entities and providing briefings to senior U.S. officials. In a parallel effort, OpenAI has announced that it will make its most advanced AI models available to vetted government bodies, aiming to preemptively tackle AI-related threats. Jessica Ji, a senior research analyst at Georgetown’s Center for Security and Emerging Technology, noted that partnerships like these could enhance CAISI's testing capabilities by leveraging resources that the agency may lack compared to large tech firms. Meanwhile, the White House is consulting with experts to explore a possible review process for new AI models, marking a shift from the previous administration's more lenient regulatory stance. The New York Times reported on the establishment of this working group, while a White House spokesperson stated that any policy announcements will come directly from the President, cautioning against speculation about potential executive orders. As the landscape of AI continues to evolve, Microsoft's Chief Responsible AI Officer Natasha Crampton highlighted that CAISI provides essential expertise in technical, scientific, and national security matters, complementing Microsoft’s internal testing efforts. While Google has opted not to comment on the agreement, xAI has yet to respond to inquiries regarding their participation.