AI, Hegseth threatens Anthropic with cancellation of Pentagon contract

Defence Secretary Pete Hegseth has set an ultimatum for Anthropic CEO Dario Amodei on Friday to accept demands to remove protections from its artificial intelligence model, or risk losing his contract with the Pentagon. Hegseth also threatened to put the AI company on a government blacklist.

The problem at the centre of the tug-of-war between Hegseth and Amodei are the barriers that Anthropic has imposed on its artificial intelligence model Claude.

The Pentagon, which has a $200 million contract with the company, wants Anthropic to lift restrictions preventing the US military from using the model for 'any lawful use', according to two sources familiar with the discussions told the New York Times.

But Anthropic has concerns about two issues it is unwilling to ignore: artificial intelligence-controlled weapons and mass domestic surveillance of US citizens.

According to a source familiar with the matter, Anthropic believes that:
1) artificial intelligence is not reliable enough to handle weapons;
2) there are still no laws or regulations governing its use in mass surveillance.