The Trump administration is reportedly preparing a major policy reversal that would bring a new era of government oversight to the AI industry. Despite previously pledging to avoid “foolish rules” that might stifle innovation, the White House is now drafting an executive order to establish a formal review process for advanced AI models before they are released to the public. +1

The catalyst for this shift is Mythos, a new model from Anthropic so powerful at identifying and exploiting cybersecurity flaws that the company itself has been afraid to release it.

Key developments in the administration’s policy shift:

• The “Mythos” Threat: Anthropic’s Mythos model has demonstrated a terrifying ability to autonomously find “zero-day” vulnerabilities across every major operating system. Anthropic withheld the model from the public, warning it could trigger a global cybersecurity crisis, which caught the immediate attention of the NSA and national security officials. +1
• A “Pre-Release” Vetting Process: The proposed executive order would create a working group of tech executives and government agencies (including the NSA and DNI) to evaluate “frontier” AI models. The goal is to give the government “first access” to test these systems for national security risks without necessarily blocking their commercial debut. +1
• Retreat from Total Deregulation: This move marks a significant departure from President Trump’s January 2025 decision to scrap Biden-era AI safety rules. While the administration still aims to promote American AI leadership, officials are increasingly worried that an AI-enabled cyberattack could cause catastrophic political and economic fallout. +1
• The Anthropic Paradox: The shift comes amid a bizarre legal and political feud. The Pentagon recently labeled Anthropic a “supply chain risk” after the company refused to grant the military unrestricted, “safeguard-free” access to its models. Now, the White House find itself needing the very technology it has been fighting in court. +1
• JD Vance’s Outreach: Vice President JD Vance recently held a high-level call with CEOs from OpenAI, Anthropic, Google, and Microsoft. While Vance has warned against “excessive regulation,” he emphasized to the tech leaders that “we all need to work together” to manage the risks posed by this new generation of autonomous hacking tools.

The proposed framework is reportedly inspired by the United Kingdom’s AI safety model, where government agencies assess whether systems meet specific security standards before they hit the market. If signed, the order would represent the most interventionist stance the Trump administration has taken on the tech sector to date.