Anthropic CEO Dario Amodei has responded to reports that the company's Claude AI may have played a role in a military strike on a school in Iran, saying he simply doesn't know whether the claims are true.
According to Business Standard, Amodei stated that even if Claude was involved, "the use case in this instance didn't violate the company's policies." He also offered a broader observation about armed conflict, arguing that "military decision makers make terrible mistakes even at the best of times."
The statement comes as the ongoing conflict with Iran has raised sharp questions about the role of artificial intelligence in modern warfare. The Irregular Warfare Initiative has framed the broader situation as a question of whether the Iran war is being fought "with or against" the AI sector — a distinction that matters as commercial AI tools increasingly overlap with military applications.
Amodei's response stops well short of a denial, and his defense rests on policy compliance rather than a factual rebuttal. That framing — that military misuse is acceptable so long as it technically fits within terms of service — is likely to intensify scrutiny of how AI companies draw the line between civilian and defense use cases.
The episode matters because it may be the clearest test yet of whether commercial AI developers can remain neutral bystanders once their products enter active war zones.