Anthropic chief Dario Amodei expressed uncertainty to Bloomberg on Wednesday about how the Claude AI model played a role in a deadly missile strike that reportedly killed around 120 children at an Iranian elementary school in late February, raising questions about the use of increasingly capable AI systems during military operations in Iran.
Amodei made the comments during Bloomberg’s "The Circuit with Emily Chang."
Photographer: Prakash Singh/Bloomberg
“Look, we don’t have access to, we don’t know exactly how these models were used,” Amodei said in an interview with Bloomberg.
The Anthropic CEO also said the use case in the instance of the school strike did not violate the company’s policies and that military leaders make mistakes “even at the best of times,” Bloomberg reported.
“The principle that we have established, and I think the principle that was obeyed here, is a human makes the final decision,” Amodei noted.
The Pentagon has not claimed responsibility for the strike as a preliminary military report found the U.S. was responsible for the bombing, according to The New York Times, which cited unnamed U.S. officials who emphasized the findings were preliminary.
An unnamed U.S. official also told The Wall Street Journal that U.S. military investigators think American forces were likely responsible for the bombing.
Forbes has reached out to the Pentagon for comment.
Amodei defended the use of AI in military conflict during his interview with Bloomberg, saying, “We don’t want a world where China and Russia can build, can analyze all the intelligence with AI, can use AI for attacking Taiwan and Ukraine, and we can’t defend them.”
President Donald Trump accused Iran of carrying out the strike “because they’re very inaccurate with their munitions, they have no accuracy whatsoever.” He later said he did not “know enough” about the bombing, which involved a Tomahawk missile—a munition not owned or used by the Iranian and Israeli militaries.
Anthropic’s AI models have been “extensively” used by the Defense Department and other national security agencies, according to the firm, which says its AI has been used for “intelligence analysis, modeling and simulation, operational planning, cyber operations, and more.” However, the relationship between the Defense Department and Anthropic hit a wall some months ago when Amodei refused a request from Defense Secretary Pete Hegseth to give the U.S. military unrestricted use of Claude. Amodei cited concerns around the potential for mass domestic surveillance and the use of fully autonomous weapons on the battlefield. The U.S. military scrapped its contract with Anthropic in response and designated Anthropic as a supply chain risk to national security. Anthropic is fighting the designation in court.
Hegseth Designates Anthropic As Supply Chain Risk After Trump Bans Government Use (Forbes)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。