Anthropic, valued at $380 billion, has imposed restrictions on use of its technology so it is not used for surveillance in the United States or for deadly violence with "autonomous weapons."
Anthropic is best known for the AI tool Claude, which was used in the attack on Venezuela in January.
Hegseth said at a meeting on Tuesday that the company must allow unrestricted use of its technology in all areas of defense. A senior Pentagon official told the newspaper that Anthropic has been given until Friday to comply with the requirement.
