Vice President J.D. Vance convened a private call in April with several of the nation's leading technology executives, cautioning them about the potential for advanced artificial intelligence systems to pose significant threats to critical U.S. infrastructure, including banks, hospitals, and water facilities. Reports published Thursday detailed the Vice President's concerns and the Trump administration's evolving approach to AI regulation.
The call included prominent figures such as Elon Musk, Sam Altman, Dario Amodei, Google CEO Sundar Pichai, and Microsoft CEO Satya Nadella. Vice President Vance reportedly urged cooperation among these industry leaders as the White House weighs potential safeguards for increasingly powerful AI systems. As a central figure in President Trump's technology policy efforts, Vance's engagement underscores a growing recognition of AI's dual-use potential.
A focal point of the discussion was "Mythos," a powerful AI model developed by Anthropic. Reports indicate that Anthropic deliberately withheld Mythos from public release due to fears it could be weaponized to attack critical infrastructure. During testing, Mythos allegedly demonstrated an advanced ability to autonomously identify cybersecurity vulnerabilities. This included reportedly uncovering a decades-old flaw in OpenBSD, which had not been previously identified by human researchers. The AI model also reportedly found weaknesses within the Linux kernel, a foundational component supporting much of the world's server infrastructure.
Officials expressed particular concern that smaller institutions, lacking the extensive cybersecurity resources of larger entities, could struggle to defend themselves against sophisticated cyberattacks orchestrated or augmented by such advanced AI systems. This apprehension prompted Treasury Secretary Scott Bessent to organize a separate private meeting last month in Washington with executives from several major U.S. banks. Attendees at this meeting reportedly included Citigroup CEO Jane Fraser, Morgan Stanley CEO Ted Pick, Bank of America CEO Brian Moynihan, Wells Fargo CEO Charlie Scharf, and Goldman Sachs CEO David Solomon, indicating the high-level concern across financial sectors.
In response to these developments, the White House is reportedly considering an executive order aimed at establishing an oversight process for advanced AI models. Furthermore, Trump administration officials have reportedly engaged Anthropic, requesting the company to limit further expansion of Mythos access specifically to organizations involved in critical digital infrastructure. Anthropic reportedly restricted the system's availability to approximately 40 companies through a confidential initiative known as "Project Glasswing." Participants in this initiative allegedly included major tech and financial institutions such as Apple, Microsoft, Google, JPMorgan, Goldman Sachs, Bank of America, Citigroup, and Morgan Stanley.
Despite these restrictions and the confidential nature of Project Glasswing, Bloomberg reported that unauthorized users managed to access the Mythos system through a third-party vendor environment during the same week as Vice President Vance's call with technology executives. These individuals reportedly located the system by leveraging knowledge of Anthropic's prior model deployment methods. The unauthorized users told Bloomberg that their intention was to experiment with the technology rather than to cause damage. Anthropic has stated it is investigating claims involving unauthorized access to the system via a vendor environment connected to its infrastructure.
These developments signal a notable shift for the Trump administration. Previously, the administration had often emphasized the need for fewer AI restrictions to maintain the U.S.'s competitive edge against China in the global race for technological dominance. This evolving stance on AI regulation comes as President Trump is scheduled to travel to China next week for meetings with President Xi Jinping, where artificial intelligence and tensions surrounding Taiwan are expected to be major topics of discussion. The discussions and potential policy changes highlight the increasing complexity of balancing technological innovation with national security and public safety in the age of advanced AI.
