Anthropic sent a letter to the U.S. Senate Committee on Banking, Housing, and Urban Affairs on June 10 accusing Alibaba of conducting what the company calls the largest known distillation attack against its models to date. CNBC confirmed the letter, which was first reported by Bloomberg.
Distillation attacks work by querying a stronger model at scale and using its outputs to train a cheaper, smaller one. The attacker gets a model that approximates frontier capability without paying frontier training costs or licensing fees. Anthropic says operators affiliated with Alibaba and the Chinese company’s AI lab ran 28.8 million exchanges with its models between April 22 and June 5, using roughly 25,000 fraudulent accounts.
The letter was addressed to Sen. Tim Scott (R-S.C.) and Sen. Elizabeth Warren (D-Mass.). Anthropic wrote that Alibaba had “brazenly” and “illicitly” attempted to extract its AI capabilities, and that the campaign proceeded despite warnings from the Trump administration. “We believe combating the threat of illicit distillation requires coordinated action between government and industry, and we will continue working with Congress and the Administration to maintain American AI leadership,” an Anthropic spokesperson said in a statement. A representative for Alibaba did not respond to CNBC’s request for comment before publication.
This is not Anthropic’s first public claim about industrial-scale distillation. In February, the company announced it had identified three separate campaigns from DeepSeek, Moonshot, and MiniMax, calling them growing in intensity and sophistication. The Alibaba accusation puts a fourth named actor on the record and provides the most granular count yet: a specific account volume, a specific query count, and a specific date window.
The letter lands in an already complicated policy environment for Anthropic. Earlier this month, the company disclosed that it received an export control directive from the Trump administration ordering it to suspend access to its two newest models, Fable 5 and Mythos 5, to any foreign national anywhere in the world, including foreign national Anthropic employees. The government cited national security authorities but did not specify its concern. Senior Anthropic staff flew to Washington to meet with administration officials; as of CNBC’s report, the company said both parties are working to resolve the issue but had not set a timeline.
That context matters for reading the Senate letter. Anthropic is simultaneously fighting the Alibaba accusation in one direction and negotiating the Fable 5 and Mythos 5 access restrictions in another. The company’s public posture, lobbying Congress on distillation threats while asking the administration to relax export controls, reflects a compressed version of the broader US-China AI competition playing out in regulatory channels.
What the announcement does not disclose is also notable. Anthropic has not said whether its fraud detection caught and blocked any portion of the 28.8 million exchanges in real time, or whether the full volume completed before the accounts were suspended. The letter does not specify what model or model version was queried, what capability Alibaba was attempting to replicate, or whether Anthropic has quantified any degradation to Claude’s competitive position. The White House memorandum from April pledged coordination against industrial-scale distillation but did not create a legal framework for enforcement; the Senate letter appears designed, in part, to build pressure for one.
For teams currently building products on top of Claude’s API, the operational question is access continuity. If the ongoing Fable 5 and Mythos 5 access dispute is not resolved quickly, those teams face a narrower model menu than they had sixty days ago. The Alibaba accusation adds a second variable: if Congress moves toward tighter API-level controls as an anti-distillation measure, usage terms for non-US companies and cross-border deployments could tighten further before the end of the year.
Reported by CNBC on June 24, 2026, with reporting contributed by Kate Rooney. Bloomberg was first to report the letter.
