OpenAI has reached an agreement to assist in artificial intelligence systems with the Pentagon after a similar deal with rival AI company Anthropic fell through. Anthropic appeared to be making strides in a $200 million contract with the Department of Defense, but negotiators could not come to an agreement on one issue: AI surveillance. Anthropic was ardently opposed to the use of its technologies for mass surveillance of the public, along with programming fully autonomous weapons. As a result of this obstacle, Defense Secretary Pete Hegseth declared Anthropic as a “supply chain risk” and ousted it from being used by government agencies. All federal agencies currently working with Anthropic have been ordered to phase out their usage over the next six months.
This did not prove to be a large setback for the DOD, as they had already been negotiating with Sam Altman, CEO of OpenAI, and announced their deal soon after Anthropic was forced out. OpenAI has expressed similar concerns regarding lawful use of their AI technologies by the government, but said that they were confident that the DOD will be ethical in their AI usage.
Despite this, Altman later acknowledged on X that their agreement had been rushed and should have been more carefully crafted. Altman emphasized that OpenAI was entirely opposed to the usage of their AI tools for domestic surveillance and intelligence purposes. He assured that these principles would be reflected in their defense deal, with cooperation from the DoD.
Anthropic is likely going to dispute its security risk designation in court. In the past, the government has only named companies from United States’ opponents as security risks, so this decision came as a shock to Anthropic and legal experts. In response to an unwillingness to limit unethical AI uses from the DoD, Anthropic CEO Dario Amodei expressed in a statement that his company “cannot in good conscience accede to their request.” However, companies like OpenAI have affirmed their faith in the United States government to follow the law and use their AI technologies responsibly.
President Trump sent out a statement on Truth Social discussing this conflict, labeling Anthropic as a “RADICAL LEFT, WOKE COMPANY.” The President later accused Anthropic of putting American lives at risk by not giving the government complete control over their AI systems. Anthropic supporters have countered by claiming that potential AI usage for surveillance and autonomous weaponry is too much of a concern to ignore, and that it would be more ethical to stick to their moral boundary.
The government has been increasingly interested in working with AI companies. Last week, Elon Musk’s xAI was approved for classified settings. OpenAI, xAI and Google have all made agreements with the government, consenting to their tools being used in government systems, as long as they remain lawful. Since the summer, Anthropic had actually been quite invested in government programs, which adds to the confusion surrounding the Pentagon’s sudden split with the company.