Trump administration discusses Washington control of AI
Updated
Updated · The Wall Street Journal · May 8
Trump administration discusses Washington control of AI
13 articles · Updated · The Wall Street Journal · May 8
The White House is also weighing pressure on tech companies to buy from government-backed vendors, widening concerns beyond earlier talks on oversight of powerful models.
The discussions suggest a sharper interventionist turn in Silicon Valley after the administration had previously pledged lighter AI regulation and now faces industry fears about competition and innovation.
Previous reports said officials were considering intelligence-community review of advanced models and a possible executive order, amid public anxiety over AI risks and debate over security versus private enterprise.
With AI discovering security flaws faster than humans can fix them, how can a global cybersecurity crisis be prevented?
As AI automates hacking, are we entering a new era of perpetual and unmanageable cyber warfare?
From Innovation to Security: The U.S. Federal AI Review Framework and the Impact of Claude Mythos in 2026
Overview
In May 2026, the Trump administration reversed its previous hands-off stance and launched formal oversight of advanced AI models, driven largely by the emergence of Anthropic's Claude Mythos in April 2026. Mythos demonstrated unprecedented ability to autonomously find and exploit software vulnerabilities, raising serious national security concerns. This led to Anthropic restricting Mythos's release and the establishment of Project Glasswing, providing controlled access to the model for defensive cybersecurity. The administration also expanded CAISI's role in pre-deployment AI evaluations and planned a cross-sector AI working group. These actions aim to balance innovation with safety amid growing threats, including AI-enabled fraud and intensified geopolitical challenges like China's disinformation campaigns.