Google, Microsoft, xAI Agree to US Government Oversight of New AI Models

The agreements bring the three companies into an existing federal review framework already used by OpenAI and Anthropic PBC, according to Bloomberg. The US Commerce Department’s Center for AI Standards and Innovation announced Tuesday that OpenAI and Anthropic have also revised their earlier arrangements with the agency to better reflect the priorities laid out in President Donald Trump’s AI Action Plan.

Federal officials said the center has been conducting pre-release reviews of advanced AI systems from OpenAI and Anthropic since 2024. During that period, the office has completed more than 40 evaluations, including assessments of leading-edge models that have not yet been made publicly available, according to Bloomberg.

The expanded agreements come as Anthropic’s Mythos system has raised concerns among US policymakers, prompting broader discussions about the role of government oversight in artificial intelligence. The center, originally created in 2023 under President Joe Biden as the AI Safety Institute, was later reorganized and renamed by the Trump administration, according to Bloomberg.

The agency now describes itself as the federal government’s main point of engagement with the AI industry for testing, collaborative research and the development of best practices. While the center remains active, its long-term legal status has not yet been formally established by Congress, though lawmakers have introduced proposals aimed at securing its future, per Bloomberg.

“These expanded industry collaborations help us scale our work in the public interest at a critical moment,” Chris Fall, the center’s director, said in connection with the new agreements.

Read more: Microsoft Ends Exclusive Access to OpenAI Technology, Opening Door to Rivals

Fall assumed leadership after the abrupt departure of Collin Burns, a former Anthropic researcher who had initially been selected for the role but left shortly after taking the position, according to Bloomberg.

The latest evaluation agreements also follow recent reports from The New York Times and The Wall Street Journal that the Trump administration is weighing an executive order that could establish a formal government review process for AI tools. Such a move would mark a significant step toward federal oversight of advanced AI systems, although a White House official described reports about potential executive action as speculative and said any policy announcement would come directly from Trump, according to Bloomberg.

Trump’s AI Action Plan, released in July, specifically directs the Center for AI Standards and Innovation to play a central role in a national AI evaluation ecosystem, including security-related assessments. The plan also encourages regulators to examine how evaluations of AI systems might fit into enforcement of existing laws, per Bloomberg.

The administration’s AI policy efforts have intensified following Anthropic’s announcement last month that its Mythos model demonstrated an unusual ability to identify weaknesses in cybersecurity defenses. Senior officials, including White House Chief of Staff Susie Wiles, Treasury Secretary Scott Bessent and National Cyber Director Sean Cairncross, have reportedly become more involved in discussions surrounding the model, according to Bloomberg.

At the same time, tensions between Anthropic and the US government continue to grow. The White House has already pushed back against efforts to expand access to the Mythos system, while the Pentagon remains locked in legal disputes with Anthropic over whether the company can be designated a supply-chain risk, according to Bloomberg.