Anthropic Warns Rivals Harvest AI Outputs at Scale, Threatening Billions in Model Spending
Updated
Updated · Business Insider · Jul 12
Anthropic Warns Rivals Harvest AI Outputs at Scale, Threatening Billions in Model Spending
3 articles · Updated · Business Insider · Jul 12
Summary
Anthropic says competitors are collecting its model outputs at scale and using them to improve rival AI systems, a practice it labels “distillation attacks.”
OpenAI and Google have issued similar warnings, arguing the tactic could let rivals replicate expensive model capabilities for a fraction of the billions spent on development.
Anthropic has spent months tightening access to its top models, but researchers say those defenses may simply push attackers toward more elaborate workarounds in a continuing cat-and-mouse game.
The dispute also exposes a broader contradiction: AI companies that defended scraping public web content as fair use are now objecting when others reuse their own outputs, with some researchers warning the backlash could chill legitimate distillation.
When rivals can copy models for a fraction of the cost, is the entire business model for frontier AI at risk?
After scraping the web for years, are AI giants now facing their own 'fair use' dilemma with model distillation?
As Trump prepares to meet Xi, can diplomacy stop AI distillation or are stricter tech controls the only real answer?
Large-Scale AI Model Distillation Attacks: U.S.-China Tensions, Economic Stakes, and the New AI Cold War
Overview
U.S. artificial intelligence companies are raising alarms about large-scale intellectual property theft, accusing Chinese competitors of copying advanced AI models through distillation attacks. This concern intensified after Alibaba banned Anthropic’s AI tools, citing security risks, and follows earlier accusations by Anthropic against Chinese firms like DeepSeek, MiniMax, and Moonshot AI for large-scale model extraction that requires advanced chips. OpenAI has also made similar claims. These incidents highlight a growing pattern where U.S. companies believe their AI innovations are being unfairly replicated, fueling tensions and prompting calls for stronger protections and government action.