OpenAI chief executive Sam Altman has admitted the company mishandled the rollout of its Pentagon deal. The AI leader took to his X (formerly Twitter) handle and admitted that the announcement ‘looked opportunistic and sloppy’ amid widespread backlash.
The criticism followed OpenAI’s disclosure that its AI models would be deployed inside classified US Department of Defense networks. Soon after, many users cancelled ChatGPT subscriptions, and the ‘Cancel ChatGPT’ trend gained traction online.
The announcement came as the US government remains in a public dispute with rival Anthropic over the use of AI systems in the military. The core disagreements involve fully autonomous weapons and domestic mass surveillance.
Anthropic has pushed for tighter limits. Its Claude app climbed Apple’s download rankings during the controversy, reflecting user support.
Altman’s latest post on X is being viewed as an attempt to calm concerns.
Altman said OpenAI is adding clearer language to its agreement with the Pentagon. The revised contract will state that it complies with the US laws, including the Fourth Amendment, the National Security Act of 1947, and the Foreign Intelligence Surveillance Act of 1978.
He said the systems will not be intentionally used for domestic surveillance of US citizens or nationals. He also clarified that intelligence agencies such as the NSA cannot use OpenAI’s models under the current contract. Any such access would require a separate agreement.
Altman called the protection of civil liberties ‘critical’. He said democratic processes must remain in control and that governments, not private companies, should make key societal decisions. He added that he would refuse to follow an unconstitutional order.
Also Read: ChatGPT Downloads Plunge After Pentagon Deal, Sam Altman Reacts
Altman acknowledged that the company moved too quickly in announcing the deal. He said the issues are complex and require clearer communication.
He described the Pentagon agreement as one of the first high-stakes decisions OpenAI has faced. He also said AI technology is not ready for every use case and that safety trade-offs need careful evaluation.
The episode highlights growing tension between AI firms, governments, and users over how powerful models should be deployed.