Pentagon Rejects AI Ethics, Paving Way for Autonomous Weapons | Human Rights Watch Warns of Risks

The escalating tension between the United States Department of Defense and Anthropic, an artificial intelligence safety firm, underscores a dangerous trajectory toward the deployment of fully autonomous weapons systems. The Pentagon’s decision on February 27th to designate Anthropic a “supply chain risk” after the company refused to cede control over ethical boundaries in its AI technology signals a willingness to prioritize military advantage over safeguards against unintended consequences. This move, coupled with a swift agreement with OpenAI, which accepted “any lawful utilize” terms for its AI products, raises serious concerns about the future of responsible AI development and its application in warfare. The core of the dispute centers on the definition of “responsible AI” and the potential for machines to make life-or-death decisions without meaningful human oversight.

A Shift in Doctrine: “Military AI Dominance”

The Pentagon’s stance represents a significant shift in AI strategy, as outlined in its January 2026 Artificial Intelligence Strategy document. The strategy prioritizes achieving “Military AI Dominance,” a goal that appears to supersede previous requirements for “appropriate levels of human judgment” in the use of force by autonomous weapons systems. This change, according to reporting from TechCrunch, effectively removes a critical check on the potential for algorithmic errors or unintended escalation. The decision to pursue AI dominance, even at the expense of ethical considerations, reflects a growing belief within the Department of Defense that maintaining a technological edge is paramount in an increasingly competitive global landscape.

Anthropic’s Red Line and the Fallout

Anthropic’s refusal to allow the Pentagon unfettered access to its Claude AI model stemmed from its commitment to preventing the development of fully autonomous weapons – systems capable of selecting and engaging targets without human intervention. In a public statement, Anthropic explicitly drew a red line at this application of its technology. This principled stand, however, triggered a swift and forceful response from the Trump administration. Defense Secretary Pete Hegseth not only moved to designate Anthropic as a supply chain risk, effectively barring it from future government contracts, but also quickly secured a deal with OpenAI, a competitor willing to accept the Pentagon’s broader terms of use. This rapid sequence of events demonstrates the administration’s determination to secure access to advanced AI capabilities, regardless of ethical concerns. President Trump subsequently ordered all U.S. Agencies to cease using Anthropic technology, as reported by the Associated Press.

The UN Convention on Conventional Weapons: A Critical Forum

The implications of this dispute extend far beyond the immediate relationship between the Pentagon and Anthropic. Governments currently meeting at the United Nations in Geneva under the auspices of the Convention on Conventional Weapons (CCW) are grappling with the challenges posed by autonomous weapons systems. The CCW provides a crucial forum for international discussion and potential regulation of these technologies. The United States’ actions, however, undermine efforts to establish meaningful safeguards. Human Rights Watch has consistently warned about the dangers of autonomous weapons, highlighting their potential to cause disproportionate harm to civilians due to their inability to distinguish between combatants and non-combatants, particularly in complex environments like protests or urban warfare.

Opacity, Accountability, and Bias: The Core Concerns

Beyond the immediate risk of unintended escalation, the deployment of autonomous weapons systems raises fundamental questions about accountability and bias. The inherent opacity and unpredictability of these systems make it challenging to assign responsibility for errors or unintended consequences. The algorithms that power these weapons are susceptible to biases, potentially leading to discriminatory outcomes and disproportionate harm to vulnerable populations, including people of color, women, and people with disabilities. These concerns are amplified by the lack of transparency surrounding the development and deployment of these technologies. The “Stop Killer Robots” campaign has extensively documented these risks, emphasizing the need for international regulation to prevent the proliferation of autonomous weapons.

A History of AI Integration in Defense

The Pentagon’s pursuit of AI integration isn’t new. As early as November 2024, Anthropic partnered with Palantir and Amazon Web Services to deploy Claude on classified military and intelligence networks, achieving a security clearance level (DISA Impact Level 6) previously unattained by other AI companies. By June 2025, Claude Gov, a version tailored for government use, was launched, followed by a $200 million contract in July 2025 to prototype frontier AI capabilities for national security. This initial integration, however, occurred under a different set of ethical constraints, which the Pentagon now appears willing to discard in its pursuit of “Military AI Dominance.” The speed with which the situation unfolded – from partnership to blacklist to replacement – underscores the high stakes and the Pentagon’s determination to maintain access to cutting-edge AI technology.

What Remains Unclear

While the immediate consequences of the Pentagon’s actions are clear – Anthropic’s exclusion from future contracts and OpenAI’s increased involvement – several key questions remain unanswered. The full extent of OpenAI’s agreement with the Pentagon, and the specific applications of its AI technology, are not yet publicly known. The long-term impact of removing the “human judgment” requirement from autonomous weapons systems remains to be seen. It is also unclear whether the United States’ actions will embolden other nations to pursue autonomous weapons development without adequate safeguards, potentially leading to a global arms race in AI-powered weaponry.

Looking ahead, the upcoming discussions at the Convention on Conventional Weapons will be critical. Governments will face pressure to strengthen the draft treaty banning and regulating autonomous weapons systems, but the United States’ recent actions cast doubt on its willingness to support such measures. The situation highlights the urgent need for international cooperation to establish clear ethical and legal frameworks for the development and deployment of AI in military applications. The path forward requires a delicate balance between national security interests and the imperative to protect human rights and prevent unintended consequences. The coming months will reveal whether the international community can rise to this challenge.