02/28/2026, 06:4402/28/2026, 06:44
After the dispute between the Pentagon and the AI company Anthropic, rival OpenAI says it has reached an agreement with the US Department of Defense on the use of its software in the military. “Tonight we reached an agreement with the War Department to deploy our models in their classified network,” wrote OpenAI boss Sam Altman on the X platform. Pentagon chief Pete Hegseth reposted the statement, as did senior official Emil Michael.
Pentagon chief Pete HegsethImage: keystone
Altman further explained that two of OpenAI’s key security principles are the prohibition of domestic mass surveillance and human responsibility for the use of force, including autonomous weapon systems. “The War Department agrees with these principles, enshrines them in law and policy, and we have incorporated them into our agreement.” Further details of the deal were initially not known.
Dispute between Pentagon and Anthropic escalates
OpenAI competitor Anthropic also insisted on the two security principles. According to its boss Dario Amodei, the company wanted to enforce limits on the use of its AI software in the US military so that the company’s AI would not be used for mass surveillance in the US or in fully autonomous weapons.
Anthropic boss Dario Amodei. Image: imago-images.de
The dispute escalated on Friday: Hegseth ordered Anthropic to be classified as a “supply chain risk to national security”. Effective immediately, US military contractors, suppliers or partners will no longer be allowed to do business with the company, Hegseth wrote at
The measure came in connection with an order from President Donald Trump that all federal agencies stop using Anthropic technology.
ChatGPT operator OpenAI and the company Anthropic are competing to accommodate their software with artificial intelligence in companies and authorities.
Altman: A level playing field for all AI companies
In his statement on The desire for de-escalation was expressed.
Sam Altman.Image: keystone
He also explained that OpenAI will develop technical safeguards to ensure that the models behave as intended, which is also in line with the ministry’s wishes. (sda/dpa)