Washington Tightens Grip as Anthropic and OpenAI Face New AI Rollout Hurdles

The Trump administration is enforcing a cautious, customer-by-customer vetting process for advanced AI models, marking a new era of state-led oversight.

The White House is no longer a passive observer in the race for artificial intelligence supremacy. In a move that signals a significant shift in how frontier models reach the public, the US government has authorised a limited rollout of Anthropic’s potent Mythos 5 model, but only for a small group of vetted cybersecurity firms and infrastructure providers. This partial reversal follows a suspension ordered on June 12, when officials flagged critical vulnerabilities in the system’s safeguards.

US Commerce Secretary Howard Lutnick confirmed the decision, noting that Anthropic has since addressed the risks that led to the initial freeze. For now, access remains restricted to "trusted partners" who play a role in national security. While this is a win for the company, it is a clear indicator that the days of unchecked product launches are over. The government is now actively shaping the deployment timeline to ensure that high-powered technology isn't turned into a weapon for cyberattacks.

The OpenAI Parallel

The shadow of government oversight is stretching across the entire industry. As Anthropic negotiates for broader access, OpenAI is facing similar scrutiny. The company recently launched its GPT-5.6 model, but the release is far from a standard market entry. Under a new federal review process established by an executive order from President Trump, OpenAI is being forced to stagger its deployment, with validation required on a client-by-client basis.

This coordinated pressure suggests a strategic pivot in Washington. By forcing firms to slow down, the administration is attempting to manage the "dual-use" nature of these tools—where the same software that boosts productivity can just as easily be used to identify and exploit security gaps in the nation's critical infrastructure.

Why it matters

The era of "move fast and break things" has collided with the reality of national security. By demanding that companies like Anthropic and OpenAI gatekeep their own tech, the US government is effectively outsourcing a layer of regulatory enforcement to the companies themselves, while keeping final veto power. For the broader tech market, this means the pace of innovation is now tethered to Washington’s risk appetite. If this trend holds, we are likely to see a permanent shift where the most powerful models serve only a privileged tier of government-approved users before (or if) they ever reach the general public.

Looking Ahead

Discussions between Anthropic and the government are expected to continue through the weekend, with the primary objective being the eventual restoration of the smaller, less powerful Fable 5. Meanwhile, the industry is waiting to see if these staggered, restricted releases become the global standard for frontier development. For investors and developers, the message is clear: the road to a wide access release now runs directly through federal oversight offices.