Anthropic Rejects the US Department of Defense's AI Request
Originally published in Turkish. Read the original →
Anthropic rejected the US Department of Defense's requests for policy changes that would open the door to using Claude models in autonomous weapon systems and mass surveillance operations. The tension between ethical AI and national security is escalating.
Anthropic Rejects the US Department of Defense's AI Request
A striking development showing that AI is not merely a technological issue but a complex power dynamic with deep ethical, legal and geopolitical dimensions became one of the most debated stories of February 2026. Anthropic, which places the mission of developing safe and beneficial AI at the center of its corporate identity, openly and firmly rejected a policy change request from the US Department of Defense (the Pentagon).
1. What It's About: The Core of the Request
The Pentagon's request to Anthropic concerned granting certain exemptions in the company's carefully drawn usage policies. In particular, it asked for restrictions to be relaxed in two critical areas:
1. Fully Autonomous Weapon Systems: Integrating Claude models into autonomous weapon platforms that can select targets and make attack decisions without human intervention. 2. Mass Surveillance Infrastructure: Using Claude in large-scale population monitoring, profiling of individuals or mass data surveillance operations.
Both of these use cases fall into categories explicitly prohibited by Anthropic's current "Acceptable Use Policy" framework.
2. Anthropic's Response
The company stated clearly that it would not accept these policy update requests. Anthropic spokespeople and leadership emphasized that these restrictions are not just a list of rules but core pillars of its approach to responsible AI development. The company's position, in short, is this: no matter how large the commercial opportunity, contributing to the construction of systems that could potentially harm humanity is incompatible with Anthropic's reason for existing.
Behind this stance lies the company's deep commitment to its founding philosophy. Much of the team that founded Anthropic left OpenAI and built a corporate identity centered on the thesis of "safer, more ethical AI." Accepting the Pentagon's request would have meant destroying the credibility built on that foundation.
3. The Tension Between the Defense Industry and AI Companies
Anthropic's refusal is the most notable manifestation of a broader pendulum swing. Since the early 2020s, big tech and AI companies have had to debate their approach to defense contracts. In 2018, Google had to cancel a contract linking its AI products to a military drone project (Project Maven) after employee protests. Microsoft, on the other hand, took a different stance, continuing to provide strong support to military systems through Azure under the JEDI defense contract.
It is clear that not all companies in the AI world are on the same page. While defense-focused firms such as Palantir and Shield AI are deliberately trying to fill this gap, "AI safety" focused companies such as Anthropic, OpenAI and DeepMind are adopting a more cautious stance with well-defined boundaries.
4. The Legal and Ethical Framework
International Humanitarian Law (IHL) still contains important unresolved questions about the legitimacy of autonomous weapon systems. The accountability gap that arises when an AI decides whom to target without human intervention is the subject of intense debate among both legal scholars and ethics researchers. Anthropic's refusal to step onto this uncertain ground signals that it has taken on the responsibility of standing guard until that legal and ethical clarity is achieved.
5. Commercial Consequences and Corporate Reputation
Rejecting the requests of the Pentagon, one of the largest and wealthiest institutional customers in the US, means taking on a very serious commercial risk. On the other hand, the decision greatly strengthens Anthropic's brand value in the field of responsible AI. Many enterprise customers, especially those operating in European and Asian markets, see this clarity and consistency as a reason to choose it.
6. The Precedent Effect
Anthropic's stance sets an important precedent across the AI industry. Other AI companies may feel the need to reconsider how they would respond when facing similar pressure. From a broader perspective, the continued absence of a global regulatory framework for the use of AI in autonomous weapon systems indicates that this debate will remain at the center of the agenda in the years ahead.
