There’s a particular kind of fury reserved for watching someone else get credit for a principle you refused to compromise on. Dario Amodei knows that feeling well right now. Last week, Anthropic walked away from a Department of War contract rather than grant the military unrestricted access to its AI. The sticking point was blunt: Anthropic wanted the DoW to explicitly commit that it wouldn’t use Claude for domestic mass surveillance or autonomous weaponry. The DoW said no. Anthropic said goodbye.
Also read: OpenAI defended Anthropic in its feud with US govt: Here’s how
Then Sam Altman walked in and signed the deal. According to a memo reported by The Information, Amodei didn’t take that quietly. He called OpenAI’s deal “safety theater” and described Altman’s public messaging around it as “straight up lies,” accusing him of falsely “presenting himself as a peacemaker and dealmaker” in what amounts to a full-throated internal broadside from one AI CEO to another.
Also read: Anthropic CEO criticises OpenAI’s defense deal, questions safety claims
The crux of the dispute is a single phrase. Anthropic objected to the DoW’s demand that its AI be available for “any lawful use.” OpenAI’s contract, by its own admission, permits use for “all lawful purposes” – nearly identical language. OpenAI argues it separately negotiated explicit carve-outs for mass surveillance, making the distinction meaningful. Amodei’s position, reading between the lines of his memo, is that relying on what’s currently “lawful” is precisely the problem. Laws change. Administrations change. A safeguard tethered to legality is only as durable as the legal landscape around it.
It’s a reasonable concern and the public seems to largely agree. ChatGPT uninstalls jumped 295% in the days following OpenAI’s announcement, per TechCrunch, while Claude climbed to number two in the App Store. Amodei noted the irony himself, writing that the “attempted spin/gaslighting is not working very well on the general public.”
What makes this more than a corporate spat is what it reveals about the two companies’ underlying philosophies. Anthropic staked its reputation and a lucrative government contract on a principle. OpenAI found a way to say yes. Whether that makes Altman pragmatic or Amodei naive probably depends on who you think will be writing the laws in five years. For now, at least, the uninstall data suggests the public has an opinion.
Also read: Google faces lawsuit alleging Gemini AI manipulated man into suicide: Here’s what happened
