Security

Anthropic Accuses Alibaba of 'World's Largest Distillation Attack'

Editorial Team · 2026-06-25 · 6 min read

In a letter to US senators and the White House, Anthropic alleges Alibaba ran 29 million Claude exchanges through 25,000 fraudulent accounts to distill tra

Anthropic accused Alibaba of running the world's largest AI distillation attack in a letter sent to US senators and the White House. The alleged operation involved 29 million Claude exchanges channeled through 25,000 fraudulent accounts.

Distillation — using a stronger model's outputs to train a smaller model — is a well-known technique. What's unusual here is the scale and the alleged systematic evasion of Anthropic's terms of service.

If the allegations hold, they'll shape how frontier AI labs treat API access going forward. Expect tighter identity verification, per-account rate limiting, and possibly outright geographic restrictions on frontier model access.