Conflicts

Anthropic alleges Chinese labs extracted AI data; CEO to meet Defense Secretary regarding military contracts

Illustration for: Anthropic alleges Chinese labs extracted AI data; CEO to meet Defense Secretary regarding military contracts
AI-generated illustration. Visual interpretation does not represent real individuals or scenes.

Feb 24, 2026

THE BARE STORY

Anthropic has accused three China-based artificial intelligence laboratories—DeepSeek, Moonshot AI, and MiniMax—of utilizing fraudulent accounts to extract technology from its Claude chatbot. According to a report released by the U.S. firm, the labs used approximately 24,000 accounts to generate over 16 million exchanges. Anthropic alleged this activity was a "distillation" campaign, a technique intended to train less capable models on the outputs of a more advanced system to replicate its reasoning and coding abilities.

Jacob Klein, Anthropic’s head of threat intelligence, stated the company has "high confidence" in its findings based on infrastructure indicators and distinct traffic patterns. While Anthropic could not quantify the exact improvement gained by the foreign labs, Klein described the capability gains as substantial. The firm warned that such distillation efforts could allow actors to bypass safety guardrails present in the original U.S. systems. The accused laboratories did not respond to requests for comment at the time of the report.

Separately, Anthropic CEO Dario Amodei is scheduled to meet with U.S. Defense Secretary Pete Hegseth to discuss the military’s application of the startup’s technology. A senior Department of Defense official indicated the meeting aims to address how the military will use the models following stalled negotiations.

The dispute reportedly centers on the scope of usage. The Pentagon has expressed a desire to use the models for all lawful purposes, while Anthropic has requested assurances that its technology will not be used for autonomous weapons or domestic surveillance. Anthropic, which was awarded a $200 million contract with the Defense Department last year, stated through a spokesperson that it is committed to national security and is engaging in good-faith conversations to resolve the issues.

Same Facts. Different Perspectives.

Two AI models. Two viewpoints. One factual foundation.

Perspective by Claude

• Eroding Global Safety Protocols The primary danger of the Chinese labs' "distillation" campaign is not merely economic loss, but the deliberate dismantling of safety architecture. By training on Anthropic’s output to bypass established guardrails, these foreign actors risk deploying powerful systems stripped of ethical constraints. This perspective values the integrity of AI safety measures as a global public good that must be protected from reckless accelerationism.

• Resisting The Automation of War Anthropic’s refusal to grant the Pentagon carte blanche represents a crucial moral firewall against the dehumanization of conflict. By demanding assurances that its technology will not be used for autonomous weapons, the company is prioritizing human accountability over profit. This stance upholds the humanitarian principle that life-and-death decisions must never be delegated to algorithms, regardless of military convenience.

• Firewalling Against Domestic Surveillance The dispute regarding the Pentagon's request for "all lawful purposes" raises alarm bells regarding civil liberties and privacy rights. Interpreting this broad language as a potential backdoor for domestic surveillance, this framework sees Anthropic’s stipulations as a necessary defense of democratic norms. The priority here is preventing the seamless integration of commercial AI into the apparatus of state control against its own citizens.

How it may affect me

As a U.S. reader:

• The alleged data extraction by foreign labs to bypass safety guardrails suggests that the digital landscape may soon include powerful AI tools that operate without the safety measures designed to protect users from harmful content.

• The negotiation over "domestic surveillance" clauses between Anthropic and the Pentagon will decide whether government agencies are permitted to use this specific commercial AI technology to monitor the activities of the general public.

• Decisions regarding the military's access to AI for "all lawful purposes" will determine whether these systems are integrated into autonomous weaponry, influencing the degree of human oversight in future national defense operations.

Read the story at

Note: All TheBareNews content is AI-generated. For additional context, reporting, and updates, you are invited to explore the news outlets linked above.