Washington has a new front in its ongoing battles over technology and power: the ethics of artificial intelligence in military operations. OpenAI’s Pentagon deal and Anthropic’s expulsion from government contracts have transformed a corporate negotiation into a political confrontation, and the stakes — autonomous weapons, mass surveillance, and the limits of government authority over AI companies — could not be higher.
Anthropic had engaged this new front with a clear and consistent position. The company’s Claude AI would support lawful military operations but would not be deployed for autonomous weapons or mass surveillance. These were conditions the company considered baseline ethical requirements, not commercial restrictions, and it maintained them despite months of Pentagon pressure and escalating threats of commercial consequences.
The Trump administration’s response was to make the confrontation explicit and public. President Trump’s directive banning all federal use of Anthropic technology, announced on Truth Social with language designed to delegitimize the company’s ethical stance, transformed Washington’s newest front into a battle the administration was clearly willing to win by force rather than by argument.
OpenAI moved into this new political terrain with a deal that Sam Altman described as principled. He stated that the Pentagon contract includes protections against mass surveillance and autonomous weapons, mirroring exactly the conditions Anthropic had been expelled for insisting upon. He also raised $110 billion on the same night, demonstrating OpenAI’s ability to thrive commercially in a political environment that had just destroyed a competitor’s government relationships overnight.
Washington’s newest front will remain active long after this week’s headlines fade. The questions it raises — whether AI companies can maintain ethical limits against government pressure, whether autonomous weapons represent an acceptable use of AI, whether mass surveillance of civilians can ever be justified — are among the most consequential the technology industry has ever faced. Anthropic answered them with a firm no and lost its government contracts. OpenAI is now attempting to answer them with a deal that it claims says precisely the same thing.
