Trump's AI Policy Reversal: National Security Panic Overrides Deregulation Gospel

Two weeks ago, Pentagon CTO Emil Michael stood before CNBC cameras and declared Anthropic's Mythos model represented "a separate national security moment." The phrase carried peculiar weight from an administration that began its term by revoking every AI safety measure inherited from its predecessor.

Anthropic's Mythos model, described by the company as capable of finding thousands of critical software vulnerabilities and too dangerous for public release, has become the catalyst forcing the Trump administration to consider oversight for advanced AI models. The White House is weighing an executive order that would give the federal government a formal role in vetting all new AI models before they hit the market, creating a working group of tech executives and U.S. officials to design the oversight process.

This represents a breathtaking policy pirouette. The Trump Administration has largely positioned itself as the opposite of the Biden White House, criticising what Trump's tech policy advisors saw as overly burdensome AI safety efforts and embracing an anti-regulation approach. These discussions would represent a sharp departure from the administration's current stance as a deregulatory champion. Immediately upon taking office, the Trump administration revoked a Biden-era executive order addressing AI risks.

The sudden reversal coincides with a leadership vacuum in White House AI policy. David Sacks, who led the administration's deregulation push as AI czar, left the role in March, with White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent having since taken a more active role in shaping AI policy.

The irony cuts deeper when examined through the lens of the administration's ongoing warfare with Anthropic itself. The Pentagon declared Anthropic a supply chain risk after the company refused to sign an agreement allowing the Pentagon to use its model Claude for "all lawful purposes," insisting on banning its use for mass domestic surveillance. Less than two months ago, President Trump condemned Anthropic as an "out of control radical left" company.

Yet the very model that should theoretically intensify this conflict has instead become Anthropic's path back into federal favour. Anthropic CEO Dario Amodei met with senior members of the Trump administration to discuss Mythos at the White House just days after it was announced, including meetings with chief of staff Susie Wiles and reportedly Treasury Secretary Scott Bessent. Trump told CNBC the AI firm "tends to be on the left," but "we get along with them," adding the firm is "shaping up."

The Nuclear Moment

Pentagon tech chief Michael explained that "the Mythos issue that's being dealt with government-wide, not just at Department War, is a separate national security moment where we have to make sure that our networks are hardened up, because that model has capabilities that are particular to finding cyber vulnerabilities and patching them."

This language echoes historical moments when technology forced rapid policy recalibration. Oppenheimer's Trinity test didn't just demonstrate nuclear fission; it fundamentally altered how governments approached existential technologies. The difference here is that Mythos represents dual-use capabilities emerging from the private sector, not classified weapons programmes.

Government agencies including the Pentagon are still able to use Anthropic's models while the legal fight plays out, and the National Security Agency is even using Mythos. The administration finds itself in the absurd position of simultaneously blacklisting a company while deploying its most sensitive capabilities across classified networks.

The White House's Office of the National Cyber Director hosted two meetings last week (one with tech and cyber companies and another with tech trade groups) to discuss the broader security concerns raised by advanced AI models, including Anthropic's Mythos Preview. The office has also been discussing an AI security framework that would require the Pentagon to lead safety testing for AI deployments for federal, state and local government levels.

The Regulatory Reckoning

In a post-Mythos world, the Trump administration appears to be reevaluating its hard line against the AI safety and security measures it once shrugged off. While Trump immediately revoked the Biden administration's AI executive order on Day 1, many of the ideas that are now being considered already existed in that order.

The administration's reversal occurs as state-level AI regulation accelerates despite federal resistance. New York Governor Kathy Hochul signed amendments to the RAISE Act on March 27, 2026, shifting the framework toward transparency and reporting requirements. Most provisions of California's Transparency in Frontier AI Act became effective on January 1, 2026.

The cyber insurance market is undergoing an AI-related transformation, with many carriers increasingly conditioning coverage on the adoption of AI-specific security controls. Insurers have begun introducing "AI Security Riders" that require documented evidence of adversarial red-teaming, model-level risk assessments, and specialised safeguards as prerequisites for underwriting.

But the regulatory momentum faces an uncomfortable reality: the same corporate actors driving innovation are the ones defining the threat parameters. Some analysts have questioned whether Mythos's capabilities justify Anthropic's dramatic framing, with some studies finding that cheaper models can achieve comparable results in vulnerability discovery.

The pattern repeats across technological revolutions. Private entities create capabilities that outpace regulatory frameworks, then profit from both the innovation and the subsequent compliance infrastructure. That cooperation will evaporate if people lose confidence that their data are safe and being used responsibly, or learn that AI products are harming people.

What happens when the next Mythos emerges not from a company eager to demonstrate responsibility, but from an adversary with no interest in controlled releases? The administration's reactive posture suggests we'll discover the answer in real-time, with consequences that cannot be contained to briefing rooms or carefully staged demonstrations.

Who, exactly, is defining what constitutes an "unreasonable risk" when the definer also profits from the solution?