Anthropic slams Chinese AI firms for harvesting data from its Claude chatbot

2026-02-23 · Source: AI – SiliconANGLE · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy, AI Policy & Ethics · Depth: Intermediate, short

Summary

Anthropic PBC accuses three prominent Chinese AI firms—DeepSeek Ltd., Moonshot, and MiniMax—of illegally harvesting vast amounts of data from its Claude chatbot. The firms allegedly created thousands of fraudulent accounts to generate millions of conversations, with DeepSeek engaging in 150,000 interactions, Moonshot over 3.4 million, and MiniMax 13 million. This process, known as "distillation," is prohibited by Anthropic's terms of service, which also bar access from China. Anthropic claims these activities pose a national security risk, potentially enabling the development of advanced military weapons or mass surveillance tools, despite its own guardrails. Rival OpenAI Group PBC has made similar claims against Chinese firms, escalating its concerns to the U.S. House Select Committee on China.

Key takeaway

For CTOs and VPs of Engineering overseeing AI development, the "industrialized distillation" campaigns highlight a critical new attack surface via model APIs. Your teams must implement advanced monitoring and security controls for AI inference endpoints to prevent unauthorized data harvesting. Relying solely on product design for IP protection is insufficient; consider the interface itself as a potential exfiltration channel and prepare for coordinated industry and policy responses.

Key insights

AI frontier models accessed via APIs present a new attack surface for industrialized data distillation.

Principles

• API access enables high-volume model replication.
• Distillation can bypass traditional IP theft methods.
• Security controls are critical for model interfaces.

Method

Competitors systematically harvest model outputs through API access, using capability-targeted prompting to replicate reasoning patterns for training their own AI systems.

In practice

• Implement robust API usage monitoring.
• Strengthen terms of service enforcement.
• Explore legal avenues for data exfiltration.

Topics

• AI Data Distillation
• AI Intellectual Property
• AI Chatbots
• National Security
• Anthropic

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Engineer, AI Security Engineer, Policy Maker

Related on AIssential

• US accuses China of “industrial-scale” AI theft. China says it’s “slander.” — The US government is preparing to counter alleged industrial-scale AI intellectual property theft by Chinese entities using distillation attacks.
• The Download: water threats in Iran and AI’s impact on what entrepreneurs make — AI is rapidly reshaping industries from e-commerce to cybersecurity and robotics, while geopolitical tensions threaten critical infrastructure.
• Anthropic says Alibaba must be punished for largest Claude cloning attack — Chinese firms are allegedly conducting large-scale distillation attacks to replicate US frontier AI capabilities, raising national security concerns.
• Amazon likely voiced concerns about Anthropic AI models before US crackdown - Business Standard — Government intervention via export controls can force global AI model shutdowns over security vulnerabilities.
• Anthropic Paid $1.5B for Pirated Data — Frontier AI labs are reporting coordinated, large-scale distillation attacks by foreign entities to extract model capabilities.
• White House reins in OpenAI's GPT-5.6 — Government oversight is emerging as a critical gatekeeper for releasing powerful, "Mythos-level" AI models to the public.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI – SiliconANGLE.