OpenAI staff debate Pentagon deal's safeguards and transparency

After OpenAI finalized its agreement with the Department of Defense, current and former employees began airing their views online. The deal, confirmed by Sam Altman, drew comparisons with Anthropic, which had refused terms that might allow mass domestic surveillance or autonomous lethal weapons.

Altman later said OpenAI is working with the Pentagon to make additions to the contract amid the backlash. Boaz Barak pushed back on the notion that OpenAI weakened safeguards, arguing the contract likely offers stronger guarantees against mass surveillance and autonomous weapons than Anthropic's earlier deal.

Miles Brundage took a more skeptical view, saying employees should assume OpenAI “caved” and framed it as not caving, though he later softened his wording and noted the organisation is complex. Clive Chan said the contract contains protections against mass domestic surveillance and autonomous lethal weapons and said he is pushing internally for greater disclosure, warning he would seek termination of the deal if that proved untrue.

United States

openai, pentagon, sam altman, anthropic, boaz barak, miles brundage, clive chan, safeguards, mass surveillance, autonomous weapons