The White House wants to check powerful AI models for risks – a computer scientist explains why AI security is so difficult

According to a May 4, 2026 New York Times report, the Trump administration wants to develop a process that would require the federal government to review the safety of powerful artificial intelligence models before approving their release. The move would stand in contrast to the government’s generally anti-regulatory approach to the industry and comes as Anthropic voluntarily postpones the release of its latest AI model, Mythos.

Anthropic was concerned because while testing Mythos, the model discovered thousands of vulnerabilities in operating systems and web browsers. The implication was that if a cybercriminal or hostile foreign agent had Mythos, they could penetrate computer systems worldwide and compromise the fundamental computer code that underlies public safety, national economies and military security.

As a result, Anthropic only granted limited access to about 50 companies and organizations that manage critical infrastructure as part of its Glasswing project.