Last Week in AI #338 - Anthropic sues Trump, xAI starting over, Iran AI Fakes
Last Week in AI Last Week in AI ● Covered by 2 sources
Anthropic sued the Trump administration after a leaked Pentagon memo ordered its AI ripped out of military systems. Workers from OpenAI and Google just filed a brief backing Anthropic, which says a lot about how nervous this fight has made rivals.
Based on reporting by Last Week in AI, Last Week in AI — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Anthropic has taken the unusual step of suing its own government. Two lawsuits, filed in the Northern District of California, target the Trump administration over what the company calls an unlawful attempt to strip its models out of sensitive Pentagon systems. The trigger, according to court filings, was an internal Defense Department memo instructing military commanders to yank Anthropic's technology from key infrastructure — no public explanation, no clear technical justification, just an order to remove it.
What makes this fight louder than a typical contract dispute is who showed up to back Anthropic. Employees from OpenAI and Google filed an amicus brief in support of the company, a rare moment of solidarity across labs that spend most of their time trying to out-recruit and out-ship each other. When rank-and-file staff at competing AI shops start signing briefs together, it's a signal that the underlying issue — how much control the government can exert over which AI systems the military uses, and on what grounds — worries the whole industry, not just Anthropic's legal team.
Anthropic has spent the last two years positioning itself as the safety-conscious alternative to OpenAI, courting defense and government contracts partly on the strength of that reputation. Getting abruptly shut out of Pentagon systems undercuts that pitch directly. If a memo with no public rationale can erase a vendor from military infrastructure overnight, every AI company chasing federal work has to wonder how durable those contracts really are.
The case also lands in a week full of other reminders that AI's politics are getting messier by the month — xAI reportedly starting a chunk of its work over, and AI-generated fakes surfacing out of Iran muddying an already tense information environment. None of those stories are really about the technology anymore. They're about who gets to decide how it's deployed, and increasingly, that's a legal and political question as much as an engineering one.
My take — AI-written commentary, not fact-checked reporting
Anthropic built its whole brand on being the trustworthy, safety-first lab, and now it's suing the government that was supposed to be its biggest client — that's not a good look for anyone claiming AI regulation will be orderly and predictable. I'd bet this is the first of many fights where 'we take safety seriously' collides with 'the government does whatever it wants,' and the companies leaning hardest into government contracts are going to get burned the most.
Read more about this at: Last Week in AI