_WHAT’S HAPPENING_
The Pentagon and Anthropic, the AI company behind the chatbot Claude, severed ties after Anthropic refused approval for its technology to be used by the US government for what it alleged was mass domestic surveillance and fully autonomous weapons.
Following the dispute, Secretary of War Pete Hegseth labeled the company a “supply-chain risk to National Security,” while President Donald Trump directed that Claude AI be removed from use across all federal agencies — effectively placing it on a blacklist. Hours later, OpenAI — the company behind ChatGPT — announced it had reached its own agreement with the Pentagon.
_THE FACTS_
Pentagon Chief Technology Officer Emil Michael reportedly demanded permission to collect and process unclassified commercial data on Americans — location data, browsing histories, GPS movement, and credit card transactions.
Anthropic offered its technology to the NSA for classified use, but only on the condition that Americans’ commercial data would be off-limits. The Pentagon refused.
Secretary Hegseth summoned Anthropic CEO Dario Amodei for a tense meeting last month lasting under an hour, then delivered an ultimatum.
On February 27, Hegseth declared Anthropic a "Supply Chain Risk," a designation never before publicly applied to an American firm.
Emil Michael publicly attacked Amodei as a "liar" with a "God-complex" on social media during negotiations.
President Trump published a Truth Social post while negotiations continued— calling Anthropic "radical Left, woke" and ordering all federal agencies to drop it.
OpenAI CEO Sam Altman called Michael on February 25 and negotiated a framework within a day.
OpenAI's contract relies on existing intelligence laws, the same frameworks that justified NSA mass surveillance revealed by Edward Snowden.
_OUR TAKE_
The standoff between Hegseth and Amodei represents a real, difficult debate about whether existing law sufficiently constrains AI's power. Hegseth believes that if Congress and the courts deem something lawful, private companies have no standing to refuse, viewing Anthropic's stance as Silicon Valley imposing its liberal values on the military.
On the other hand, Amodei believes that removing safety guardrails or supporting uses of its AI that could enable mass data collection and surveillance, or autonomous drone swarms that cannot refuse “illegal orders” as human soldiers can. He argues that companies have a responsibility to set clear limits, even under government pressure.
In the global AI race, the US military must have near-unconstrained access to the most powerful technology on the planet. But if this access will infringe on Americans’ civil liberties, workarounds may need to be implemented.













