Military Use of Anthropic's Claude Raises Ethical Questions Amid Iran Strikes

Compiled by the editorial desk with reference to official statements, public filings, and industry reports.

The joint US-Israeli military campaign against Iran has resulted in 555 fatalities, including 165 deaths from an attack on an elementary school in southern Iran. As the operation unfolded, reports from the Wall Street Journal revealed that the military strike force used Anthropic's Claude chatbot to help select targets.

According to the paper, Claude, a large language model developed by Anthropic, serves as a key AI tool for US Central Command in the Middle East. Its functions include analyzing intelligence, simulating war games, and identifying military targets—essentially aiding military leaders in planning attacks that have already claimed hundreds of lives.

Anthropic's involvement in these devastating attacks may surprise those who believed the company's ethical guidelines prohibited any military work. The company and its CEO, Dario Amodei, have been in a contentious dispute with the Trump administration over two specific moral boundaries: using Claude for surveillance of US citizens and for fully autonomous lethal weaponry.

However, using Claude for target selection does not appear to violate the bot's ethical safeguards. This is notable because Anthropic has spent late February in conflict with the Pentagon over Claude's use. Last week, the Pentagon, which currently employs Claude across its classified systems, set a deadline for Anthropic to abandon those dual redlines. Anthropic allowed the deadline to pass without yielding, which many interpreted as a principled stand against the administration's militarism.

Ethical Lines Under Scrutiny

Pulitzer Prize-winning national security journalist Spencer Ackerman pointed out that Anthropic's ethical lines ignored certain issues when it initially signed its military contract. Ackerman noted that Amodei does not seem to consider building a surveillance panopticon of foreigners as a problem. He argued that the time to worry about Amodei's concerns was before signing the contract he didn't want to abandon. Ackerman lamented, "America is in such steep decline that we don't even make Oppenheimers like we used to."

Ackerman also drew a sharp analogy, saying, "When you take Doctor Doom's money to provide him a lathe to construct components for anthropomorphic robots, do you not understand that he is going to build Doombots?"

The situation highlights the complex ethical landscape for AI companies involved in military applications. While Anthropic has drawn lines on certain uses, its participation in target selection raises questions about the consistency and scope of its ethical commitments.