OpenAI's deal with the Department of War only bans "domestic mass surveillance."

Domestic.

Meaning AI-powered mass surveillance on the rest of the world is fine.

Their other safeguard? Deploying on "cloud networks only." That's not saying no to weapons or surveillance. That's saying not on our servers.

The Pentagon gets the models, the know-how, and runs it all in-house. No ethical prohibition. Just a technicality.

Anthropic refused unless the DoW promised no surveillance and no killing. They lost a $200M contract for it. OpenAI saw the opening and took it.
https://blossom.primal.net/cf27fd052711b5d78b570cd262c31df1ebe7e1fbc5b378d741969c2b5b609086.jpg