Incompatible with democratic values: American start

The Californian start-up Anthropic refused, on Thursday, February 26, to grant the US military unrestricted use of its artificial intelligence (AI), two days after the ultimatum issued by Defense Minister Pete Hegseth. “These threats do not change our position: consciously, we cannot comply with their request“, wrote Dario Amodei, leader of Anthropic, in a statement.

While the US military and intelligence agencies have already used Anthropic’s technology for US defense, it does not give the green light for two specific cases: mass surveillance of US citizens and fully autonomous lethal weapons. “Using these systems for domestic mass surveillance is incompatible with democratic values”, Dario Amodei believes.

The leader emphasizes that the most advanced AI systems are not yet reliable enough to be entrusted with the power to control lethal weapons and, therefore, to kill, without ultimate human supervision. “We will not knowingly provide a product that puts American military personnel and civilians at risk”, he concludes. “Anthropic understands” that it is the Department of Defense, “not private companies, that makes military decisions. However, in a limited number of cases, we believe that AI can undermine democratic values rather than defend them”, he states.

The Department of Defense has asked its AI suppliers to lift usage restrictions on their models by default, to expand possible use cases as long as they remain legal. All have agreed, including Anthropic, except for the two cases mentioned by its leader.