Trump Administration Considers New Oversight for Advanced AI Models

Anthropic has developed an AI model called Mythos so advanced that the company will not release it publicly, according to reports. The model can independently discover software vulnerabilities, raising concerns it could be used to trigger cyberattacks on small-town banks, hospitals and water plants.

Vice President JD Vance warned the heads of Microsoft, Alphabet, OpenAI and Anthropic on a recent call about the risks posed by Mythos. Vance, who last year told a Paris AI summit that overregulation could harm the industry, signaled the need for greater scrutiny of the most capable systems.

The warnings have prompted the Trump administration to rethink its approach to AI regulation. This marks a reversal from the administration's December 2025 order that targeted state AI laws. Officials are now moving toward a more structured federal review regime for cutting-edge systems.

The White House previously favored minimal federal intervention in AI development. That stance is shifting as new tools demonstrate the ability to expose hidden security flaws in software. One official likened the expected oversight framework to the FDA's process for approving drugs.

The comparison suggests a rigorous evaluation before the most powerful models are deployed or released. Anthropic's decision to withhold Mythos from public release underscores the company's own assessment of its risks and capabilities. The model represents a new breed of hacking tools that can operate without human direction.

Microsoft, Alphabet and OpenAI were included alongside Anthropic in the discussion led by Vance. The conversation focused on how autonomous vulnerability discovery could be misused. Such capabilities could lower the barrier for malicious actors targeting critical infrastructure.

The administration's evolving position comes as AI systems grow more sophisticated. Reports indicate Trump officials are starting to reconsider their initial regulatory restraint. The Wall Street Journal first reported details of the policy reconsideration and the concerns surrounding Mythos.

Anthropic, a privately held company, has not publicly detailed the model's full capabilities.