Scoop: Anthropic CEO apologizes for leaked memo criticizing Trump
TL;DR
The Pentagon formally designated Anthropic a supply chain risk – an unusual and far-reaching move against an AI company.
Key Points
- CEO Dario Amodei apologized for a leaked internal memo sharply criticizing the Trump administration, which had put ongoing negotiations at risk.
- Despite the apology, Anthropic still intends to sue over the Pentagon designation, which the company says narrowly restricts only certain activities.
- Amodei's initial strong stance had won him widespread praise and reportedly boosted Claude user numbers – his reversal is now drawing disappointment.
Nauti's Take
Amodei went from hero to apologizer in record time – and that is not a personal failure but a systemic one. AI companies want government money without the political compromises that come with it.
Anyone who believes you can simultaneously claim moral leadership and hold lucrative Pentagon contracts will be caught out by reality sooner or later. The tragedy: Anthropic's original position was substantively justified – the apology now undermines exactly the credibility that was supposed to set the company apart from OpenAI.
Context
The case illustrates how fragile the relationship between AI companies and the US government is: a single leaked memo was enough to derail negotiations and trigger an official Pentagon risk designation. For the entire industry, it serves as a warning that pursuing federal contracts comes with political risks that go far beyond ordinary business exposure. The deeper question of how AI may be used for military and surveillance purposes remains structurally unresolved.
Sources