Trump calls Anthropic safe—but their smartest AI is still in digital jail

The sudden pivot from the White House came just two days after Trump and Anthropic CEO Dario Amodei crossed paths at the G7 summit in France. The president, who previously labeled the company a national security threat, shifted his tone from nuclear-level alarm to a casual 'not right now.' This diplomatic thaw follows a chaotic June 12th, when Commerce Department export controls forced Anthropic to pull the plug on its most powerful models, Fable 5 and Mythos 5, effectively rendering them radioactive to anyone outside a specific US-approved bubble.

Trump defended the move by praising Amodei for his 'very fast' reaction to legal pressures, acknowledging that the tech firm was staring down a massive, career-ending liability nightmare. Despite the sudden warmth, the actual regulatory handcuffs remain firmly locked. The export controls are still active, the models stay offline, and the Pentagon hasn't retracted its earlier warnings about supply chain risks. Trump even dropped a reminder that the Defense Production Act remains a loaded gun on his desk, capable of forcing any private tech company to march to the government's beat at a moment's notice.

This is a masterclass in performative stability. While the president smooths over the optics, his team is busy hammering out a new standard for AI vulnerability, essentially trying to write a rulebook for when exactly the government gets to play god with private code. It is a fragile equilibrium where a single handshake can briefly pause a geopolitical firestorm, yet the underlying mechanisms of control remain untouched, waiting for the next political wind to change direction.

Source: CNBC