What happened

The Pentagon used Anthropic’s Claude AI model (integrated via Palantir Technologies) to assist in the planning and operational phases of a U.S. special operations forces aid on Caracas, Venezuela, to bomb military complexes and abduct President Nicolás Maduro and his wife, Cilia Flores. 

The raid resulted in 83 to 100 people killed, including security personnel and dozens of Cuban nationals, serious damage to the capital’s infrastructure, including localised power outages and a state of national emergency, and the removal of a head of state.

Why it happened

The controversy stems from the rapid, opaque integration of Claude into U.S. classified military operations. 

While Anthropic’s "Acceptable Use Policy" explicitly prohibits using Claude for violence, weapons development, or mass surveillance, the model was provided to the Pentagon through Palantir, a third-party contractor. This middleman arrangement created a "transparency fog" where the AI developer was allegedly unaware of the specific lethal mission its tool was supporting.

Anthropic positions itself as a "safety-first" company, but the Trump administration’s "AI-first warfighting" strategy prioritises speed and tactical dominance. The Pentagon argues that private companies should not be allowed to impose "ethical vetoes" on lawful military operations, leading to a direct clash between corporate safety constitutions and national security imperatives.