‘Anthropic is lying’: US defence official rebuts AI lab over military use of Claude

A sharp public exchange has erupted between artificial intelligence firm Anthropic and a senior US defence official, underscoring growing friction between frontier AI labs and national security agencies over the military deployment of advanced models.

In a post on X, Emil Michael, the US Under Secretary of War accused Anthropic of “lying” about the nature of its discussions with the Department of War. “Anthropic is lying,” Michael wrote, adding that the department does not conduct mass surveillance “as that is already illegal.” Instead, he said the goal is to allow US warfighters to use AI tools in combat scenarios without having to seek approval from Anthropic CEO Dario Amodei, including in situations involving enemy drone swarms.



The remarks followed a statement from Anthropic where Amodei addressed the company’s engagement with the Department of War.

Anthropic, which builds the Claude family of models, has positioned itself as a safety-first AI lab. Its existing government contracts reportedly include clauses preventing the use of its systems for autonomous weaponry and mass surveillance of Americans.

Community notes appended to the defence official’s post claimed that the department is seeking removal of those contractual restrictions, intensifying scrutiny over the scope of permissible AI use in defence settings.

The episode highlights a widening global debate over how far frontier AI models should be embedded into military operations. While AI companies are increasingly partnering with governments, many have retained usage policies that restrict lethal autonomous applications.

As militaries push for faster deployment of AI-enabled capabilities, the clash signals a deeper fault line between national security demands and the guardrails that AI labs argue are essential to prevent misuse and retain public trust.