clauding.de
DE
• 2 min read AI-generated

U-Turn: Trump No Longer Sees Anthropic as a Security Threat

Copy article as Markdown

A week ago, the White House treated Anthropic like an adversary — export controls, a Pentagon warning, models switched off. Now Trump tells Axios: a threat? 'Not now.' So what changed?

Anthropic Trump National Security Fable 5 Politics
Featured image for "U-Turn: Trump No Longer Sees Anthropic as a Security Threat"

That was fast. One week, the US government effectively labels Anthropic a national security threat. The next, Donald Trump says he doesn’t really see it that way anymore. In an interview for ‘The Axios Show’ on June 19, he was asked whether he viewed Anthropic as a threat. His answer: ‘Well, not now — but a week ago, maybe.‘

What Came Before

To understand the reversal, rewind a bit. The week prior, the Trump administration had blocked access to Anthropic’s most powerful models — Mythos 5 and Fable 5 — for every country outside the US and for foreign nationals inside it. The Commerce Department imposed sweeping export controls, and the Pentagon flagged Anthropic as a supply chain risk.

The trigger was apparently a report from Amazon detailing a vulnerability. The administration took it to Anthropic — and felt brushed off. Suddenly an American AI lab was being treated like a foreign adversary. For the whole industry, that was a shock.

What Flipped It

And then? The G7 summit. Trump says he came away with the impression that Dario Amodei was ‘nice’ and ‘smart.’ Sometimes it really is the personal encounters that move policy. Both sides are now reportedly working on shared standards to evaluate AI jailbreaks — exactly the security issue that sparked the fight in the first place.

The threat isn’t fully off the table, though. Trump didn’t rule out invoking the emergency powers of the Defense Production Act if Anthropic doesn’t fall in line. So the message is: peace, with conditions.

What I Think

I find this episode remarkable for two reasons. First, it shows how thin the ice is that frontier labs now operate on. One vulnerability report, one frustrating phone call — and a company with a multi-billion-dollar valuation lands under export controls. That’s an entirely new scale of political risk.

Second: if an escalation like this can be undone within days over a single summit, then clearly a lot rides on personal relationships and little on firm rules. That’s reassuring and unsettling at the same time. For Anthropic, heading toward an IPO, the relief is probably real regardless. For now.

Sources: