Anthropic accuses Alibaba of largest known distillation attack on Claude

In a 10 June 2026 letter to the US Senate Banking Committee, first reported by Bloomberg, Anthropic alleged that operators affiliated with Alibaba and its AI lab ran what it calls the largest known distillation campaign against Claude: roughly 28.8 million model exchanges through about 25,000 fraudulent accounts over a 44-day window (22 April–5 June 2026). Anthropic says the effort targeted Claude's software-engineering, agentic-reasoning and cybersecurity capabilities in the frontier model it calls Mythos Preview. Sen. Elizabeth Warren cited it as 'the largest known distillation attack on Anthropic to date.'

These are allegations, and Alibaba denies them — the 28.8-million and 25,000 figures are unverified. Distillation trains a cheaper model on a stronger one's outputs; Anthropic's worry is that the copy gets Claude's capability without its safety training, usage policies or access controls. Anthropic frames this as an escalation of a pattern: in February 2026 it named DeepSeek, Moonshot and MiniMax, and says the alleged Alibaba campaign is larger than all three combined. The disclosure also lands the same week Anthropic faced US export-control pressure — context that colours the timing.

Expect the dispute to run through Washington rather than a courtroom in the near term: the accusation is now part of the record before a Senate committee, and it feeds directly into the fight over frontier-model export controls and how the US treats its leading AI labs. Watch for any independent verification of the numbers, a fuller response from Alibaba, and whether lawmakers use the claim to justify tighter rules — none of which is settled yet.