OpenAI's fast-tracked Pentagon deal just got a civil-liberties rewrite. After a weekend of blowback, the company said Monday that its agreement to supply AI for the Defense Department's classified systems now explicitly bars using its tools for "intentional" domestic surveillance of Americans, including tracking people via commercially bought personal data, the New York Times reports. The revised terms come days after the Trump administration branded rival Anthropic a "supply-chain risk" and moved to cut it off from federal work after Anthropic refused to sign a broad "all lawful purposes" clause without bans on domestic spying and autonomous lethal weapons.
OpenAI, which did accept the "all lawful purposes" language, instead negotiated the right to embed technical guardrails and have some employees work inside the Pentagon on classified projects to oversee safety, the Times reports.
- In a post on X, CEO Sam Altman, who had publicly supported Anthropic's push against surveillance and killer robots, said: "We shouldn't have rushed to get this out on Friday. The issues are super complex, and demand clear communication. We were genuinely trying to de-escalate things and avoid a much worse outcome, but I think it just looked opportunistic and sloppy. Good learning experience for me as we face higher-stakes decisions in the future."
- He also said the Pentagon assured OpenAI that defense intelligence agencies, including the NSA, will not use its technology. Anthropic did not comment.
- Amid the backlash against OpenAI, social media users urged people to delete ChatGPT and Claude, Anthropic's rival product, rose above ChatGPT in download charts, the Guardian reports.
- Despite the order to start cutting ties with Anthropic, the military still used Claude in its attack on Iran.
,
uBlock 
,
and
Ghostery 
)
