Anthropic CEO Dario Amodei has taken a position that most tech executives would consider career suicide: telling the Pentagon “no.”
The company behind the Claude AI model has refused to comply with Department of Defense demands to strip safeguards from its technology for certain military applications, citing concerns about mass domestic surveillance and fully autonomous lethal weapons.
The red lines Anthropic won’t cross
Amodei declared on February 26 that Anthropic could not comply with the Pentagon’s requests for unrestricted Claude deployment. The company has drawn two explicit boundaries: no mass domestic surveillance and no fully autonomous weapons systems.
The reasoning isn’t purely philosophical. Amodei has argued that deploying Claude in these contexts could lead to terrible mistakes, pointing to reliability risks that current AI models still carry.
