Anthropic has accused Alibaba of orchestrating the largest known campaign to illicitly extract Claude's capabilities, utilizing nearly 25,000 fraudulent accounts to generate over 28.8 million interactions. The company claims these attacks targeted high-value features like agentic reasoning and software engineering, while allegedly ignoring previous warnings from the US administration.
Background
US AI firms are increasingly concerned about Chinese competitors using API scraping and distillation attacks to replicate frontier models without bearing the substantial costs of training them.
- Source
- Ars Technica
- Published
- Jun 26, 2026 at 02:01 AM
- Score
- 9.0 / 10