The US Department of Defense (DoD) and Anthropic are at odds over how AI should be used by the military. The disagreement has escalated to the point where Defense Secretary Pete Hegseth issued a stark warning: work with us on our terms or face being cut out from Pentagon programs. This tense standoff was highlighted in reports that Hegseth gave Anthropic until February 27 to accept the DoD’s conditions, with the threat of being labeled a supply chain risk if they refused.

Clash Over AI Use and Ethical Boundaries

The core issue revolves around how much control the DoD should have over the AI tools developed by Anthropic. The department argues that it should be able to use Anthropic’s AI for any lawful purpose, including military operations, without restrictions. In contrast, Anthropic wants to set specific boundaries to prevent misuse or unethical applications of its technology. The company’s leadership emphasizes the importance of AI safety and ethics, advocating for tighter guardrails on how their models are deployed in military contexts.

According to a report, the friction increased after Anthropic CEO Dario Amodei insisted that the DoD respect the limits the company had placed on its AI’s use in certain military activities. These disagreements came into sharper focus following a January operation where the US military used Anthropic’s Claude language model alongside Palantir’s technology to plan and execute the capture of Nicolás Maduro, the former Venezuelan president. Staff at Anthropic reportedly raised concerns internally about whether such a mission aligned with their safety and ethics guidelines.

Concerns About AI Ethics and Safety

Anthropic’s safety policies, detailed in its Acceptable Use Policy, prohibit using its AI for mass domestic surveillance, attacking critical infrastructure, or developing weapons. However, the company’s founder has also spoken publicly about the need for regulation and responsible AI use. Dario Amodei has expressed discomfort with decisions about AI being made by only a few companies or individuals, calling for more oversight and regulation to ensure the technology benefits society and avoids harm.

The disagreement has broader implications for the AI industry and government contracts. If the DoD follows through with banning Anthropic, it could impact the entire supply chain involved in military AI development. Many companies in the Defense Industrial Base rely on partnerships and technology from firms like Anthropic. A ban could force the Pentagon to seek alternative solutions, potentially slowing down military AI projects or increasing costs.

This ongoing dispute underscores the growing tension between technological innovation and ethical boundaries. As AI becomes more integrated into defense operations, questions about safety, responsibility, and control are likely to intensify. The outcome of this standoff could shape how military AI is regulated and used in the future, emphasizing the need for clear standards and responsible development practices.

Inspired by

• https://www.computerworld.com/article/4137506/us-dod-to-anthropic-compromise-ai-ethics-or-be-banished-from-supply-chain-2.html

Sources

• axios.com
• semafor.com
• cio.com
• anthropic.com
• darioamodei.com