Pentagon's Conflict with Anthropic: Department Demands Removal of Claude Model Safeguards

Tensions between the government administration and the tech sector have reached a critical point. Defense Secretary Pete Hegseth has given the company Anthropic a deadline until this Friday to abandon the internal safety restrictions imposed on the Claude model. The Pentagon demands full access to the technology to freely utilize it for military purposes. Hegseth, representing a new, more rigorous approach to civilian suppliers, described the company's current safeguards as "woke AI" that he claims weakens the United States' defense potential in the face of China's growing power. The situation is exacerbated by reports of progress from Chinese competitors. Although the U.S. administration recently eased restrictions and enabled licensed exports of Nvidia H200 processors to China, Chinese entities such as DeepSeek are still accused of earlier sanctions evasion and illegal data acquisition through intermediaries. Anthropic maintains that Chinese companies MiniMax and Moonshot used thousands of accounts to "distill" knowledge from the Claude model, allowing them to build their own systems via shortcuts. The Pentagon argues that under such conditions, American AI creators cannot afford excessive ethical brakes that slow the adoption of technology in the military. Cooperation between the Pentagon and Silicon Valley has historically been difficult, as demonstrated by Google employee protests against Project Maven in 2018, which concerned the use of AI for analyzing drone footage. Despite the threat of canceling a $200 million contract, Anthropic's leadership has so far not yielded to pressure. The company emphasizes that its mission is to build safe and controllable artificial intelligence, and that removing control mechanisms could lead to unpredictable consequences. Experts point out, however, that the startup's position is difficult, as the Pentagon is considering redirecting funds to competitors who demonstrate greater flexibility in cooperating with the defense department. „I will not allow our military to be held back by ideological safety guardrails while our adversaries have no such qualms.” — Pete Hegseth

Perspektywy mediów: Emphasizes the right of private companies to decide on the ethical use of their proprietary technologies and the risk of AI militarization. Highlights the necessity of fully subordinating technology to national security needs in order to win the race with China.