Washington – Anthropic, a leading artificial intelligence company, is engaged in a high-stakes effort to salvage its relationship with the U.S. Department of Defense after negotiations over the use of its AI models in military operations broke down last week. The dispute centers on Anthropic’s insistence on ethical guardrails, specifically preventing the use of its technology for mass surveillance and autonomous weapons systems, a stance that has drawn sharp criticism from within the Pentagon and put a $200 million contract in jeopardy. The situation highlights the growing tension between the rapid advancement of AI and the complex ethical considerations surrounding its deployment by the military.
At the heart of the current negotiations is Anthropic CEO Dario Amodei, who is reportedly in direct talks with Emil Michael, the under-secretary of defense for research and engineering, according to multiple sources familiar with the matter. The goal is to reach a revised agreement that allows the Pentagon to continue utilizing Anthropic’s AI capabilities whereas addressing the company’s concerns about potential misuse. A failure to reach a deal could result in Anthropic being designated a “supply chain risk,” effectively barring it from working with the military and its contractors.
A Clash Over Ethical Boundaries
The current impasse stems from a fundamental disagreement over the permissible uses of AI in defense. Anthropic, which developed the Claude AI model, was the first to offer its technology for use in classified settings and by national security agencies in July 2023. However, the Pentagon has sought broader access, pushing for the ability to use AI for any “lawful” purpose. This stance clashed with Anthropic’s pre-defined “red lines,” which include prohibitions against mass domestic surveillance and the development of fully autonomous weapons.
The breakdown in talks was particularly acrimonious. According to the Financial Times, Michael reportedly attacked Amodei, calling him a “liar” with a “God complex” during a Thursday meeting. Deliberations ended the following day when the Pentagon offered to accept Anthropic’s terms if the company removed a specific phrase from the contract related to the “analysis of bulk acquired data” – a clause Anthropic viewed as crucial to preventing unwanted surveillance. Amodei, in a memo to staff, described the offer as “very suspicious.”
In the memo, first reported by The Information and seen by the FT, Amodei also expressed frustration with what he characterized as misleading statements from the Pentagon and OpenAI, another AI company that recently reached an agreement with the Defense Department. He suggested that Anthropic was being penalized for not offering “dictator-style praise to Trump,” in contrast to OpenAI’s CEO Sam Altman.
OpenAI’s Agreement and the Supply Chain Risk Designation
The timing of the dispute coincides with OpenAI’s recent agreement with the Pentagon, a deal that appears to have been reached on more favorable terms for the Defense Department. This agreement, reached on Friday, has fueled speculation that Anthropic is being singled out for its unwillingness to compromise on its ethical principles.
Defense Secretary Pete Hegseth has threatened to designate Anthropic as a supply chain risk, a move that would effectively isolate the company from the military-industrial complex. This designation would compel businesses within the military supply chain to sever ties with Anthropic, significantly hindering its ability to operate in the defense sector. Amodei has vowed to challenge this designation in court, calling it “retaliatory and punitive.”
The situation is further complicated by the fact that Anthropic was involved in a controversial raid in January against Venezuelan President Nicolas Maduro. The Pentagon reportedly used Anthropic’s Claude AI model during the operation, raising concerns within the company about the potential for misuse. A top Anthropic official reportedly questioned Palantir, a defense contractor, about how Claude was used in the raid, prompting the Pentagon to review its relationship with Anthropic, as reported by Fortune.
What’s Next for Anthropic and the Pentagon?
The coming days will be critical as Amodei and Michael attempt to bridge the gap between their respective positions. The outcome of these negotiations will have significant implications for the future of AI in defense, setting a precedent for how ethical considerations will be balanced against national security interests. The Pentagon’s insistence on broad access to AI technology, coupled with its willingness to penalize companies that raise ethical concerns, raises questions about the extent to which the military is willing to prioritize ethical considerations in its pursuit of technological advantage.
As of now, Anthropic and the Pentagon have both declined to comment on the ongoing discussions. However, the stakes are high for both sides. For Anthropic, the potential loss of a $200 million contract and access to the lucrative defense market represents a significant setback. For the Pentagon, the loss of access to a leading AI model could hinder its ability to maintain a technological edge in an increasingly competitive global landscape.
The next step will likely involve further discussions between Amodei and Michael, potentially with the involvement of other senior officials from both sides. The resolution of this dispute will not only determine the fate of Anthropic’s contract but also shape the broader debate about the responsible development and deployment of artificial intelligence in the military.
Additional reporting by Joe Miller in Washington.
This is a developing story. Readers are encouraged to follow updates from official sources, such as the Department of Defense and Anthropic, for the latest information.
Worth a look
