OpenAI Promises AI Won't Spy Or Kill... Unless Pentagon Says Please
KEY POINTS
- ā¢OpenAI confirmed last week that it has a deal giving the Department of Defense access to its AI models.
- ā¢Employees on X debated fiercely whether OpenAI weakened safeguards compared to Anthropicās refusal of similar government terms.
- ā¢Following backlash, CEO Sam Altman said OpenAI is working with the Pentagon to amend the contract and clarify safety guarantees.
In a weekend saga straight out of corporate satire, OpenAI staff publicly indeclared their epic Pentagon babysitting contract. Sam Altman confirmed last week the DOD now has AI access, prompting a Twitter circus with ex-head of policy research Miles Brundage accusing OpenAI of cleverly staged surrender while pharmaceutical-grade policy wonks like Boaz Barak (also a Harvard prof) argue OpenAIās new deal actually blocks mass surveillance and killer robots better than Anthropic ever did. Meanwhile, Clive Chan demands more transparency, Mohammad Bavarian calls Pentagon paranoia 'extreme overreaction,' and Noam Brown warns democracy shouldn't be an optional AI setting. Altman later admitted they'll 'make additions,' aka patch the contractās HR hug loopholes.
Share the Story
Source: Businessinsider | Published: 3/3/2026 | Author: Lee Chong Ming