How Anthropic’s mythos threw the White House AI strategy into chaos
Vice President JD Vance raised concerns during a White House call with AI executives about Anthropic’s Mythos model, which can independently identify software vulnerabilities, potentially enabling cyberattacks on critical infrastructure like banks and hospitals. The administration is now considering an executive order to oversee advanced AI models, delaying Mythos’ wider access and sparking internal debates over regulation versus innovation.
Vice President JD Vance expressed alarm during a recent White House call with CEOs of major AI firms, including Anthropic’s Dario Amodei, OpenAI’s Sam Altman, and others, over the risks posed by Anthropic’s Mythos model. The AI system can autonomously detect software vulnerabilities, raising fears of cyberattacks on small-town banks, hospitals, and water plants—infrastructure local governments lack the tools to defend. The April meeting followed a White House briefing that heightened concerns about advanced AI capabilities, prompting a shift in the administration’s approach. Instead of focusing solely on outpacing China in AI development, officials are now weighing an executive order to establish formal oversight for the most powerful models. The White House has requested Anthropic delay expanding Mythos’ access to critical infrastructure operators while assigning National Cyber Director Sean Cairncross to lead the response. Internal divisions have emerged, with some officials and congressional aides warning of an overreaction, while AI safety advocates support stricter controls. White House adviser David Sacks, a venture capitalist, dismissed the concerns as exaggerated, arguing that proper use of AI tools could enhance digital security. Meanwhile, Chief of Staff Susie Wiles emphasized balancing oversight with innovation, framing the effort as an ‘America First’ approach to safe technology deployment. The administration remains cautious about models like OpenAI’s upcoming GPT-5.5-Cyber, which shares similar capabilities to Mythos. OpenAI reportedly consulted the White House before previewing its cyber-focused AI and has also restricted access. Discussions between the U.S. and China on AI risks are expected to begin ahead of next week’s summit in Beijing, signaling a broader geopolitical focus on managing AI threats. Any executive order would require formal announcement by President Trump, and officials stress that decisions are still under consideration. The debate reflects broader tensions between fostering AI innovation and mitigating risks, with stakeholders divided over the need for government intervention.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.