Story perspectives

Anthropic Accuses Alibaba of 28M-Interaction AI Distillation Attack

6/26/2026

Summary

Anthropic Accuses Alibaba

Anthropic sent a June 10, 2026 letter to Tim Scott (R-SC) and Elizabeth Warren (D-Mass.) accusing Alibaba of a distillation attack on Claude.
Alibaba’s Qwen AI lab used 25,000 fake accounts to generate 28.8 million Claude interactions between April 22 and June 5, 2026, targeting software-engineering, agentic-reasoning and long-horizon capabilities.
Anthropic called the campaign illicit, largest distillation attack, urged export controls, noted Alibaba’s stock fell ~3 percent and the Commerce Department restricted Mythos and Fable models.

Summary

Alibaba Sues Pentagon

Alibaba filed a June 2026 lawsuit to remove itself from the Pentagon’s 1260H blacklist, causing its shares to fall 4 percent on June 25.
Anthropic flagged its Mythos model for cybersecurity concerns because it can exploit weaknesses, after naming DeepSeek, Moonshot and MiniMax for distillation attacks on Claude.
China sanctioned ten U.S. firms linked to the military, prompting Anthropic’s leadership to call for coordinated action as the firm prepares for a public listing.

Summary

Anthropic Exposes Alibaba

Anthropic identified Alibaba’s AI unit Qwen as the division that ran 25,000 fraudulent accounts in the Claude distillation campaign.
A June 10 letter says the Alibaba operation is the largest Claude distillation effort recorded.
Alibaba shares fell 2.73% to $99.80, rose to $100.15, and were ranked by Benzinga Edge in the 94th percentile for Value.
Analysts say the high value percentile and after-hours rise show optimism despite the AI theft claims.

Summary

Anthropic Flags Alibaba

The company reported a massive unauthorized distillation campaign that harvested outputs from a more advanced AI and trained a cheaper model without source code.
Anthropic warned that models produced through this technique often lack the original system’s safety guardrails, letting rivals replicate capabilities at a fraction of development cost.
The June 10 letter describing the campaign was first reported by Bloomberg before being obtained by Reuters.

History

Historical Context

IBM’s Deep Blue defeated world champion Garry Kasparov in May 1997, becoming the first computer to win a match under tournament time controls and earning a $700,000 prize. The victory sparked early concerns about AI’s strategic power, a theme echoed in Anthropic’s June 10 2026 letter accusing Alibaba’s Qwen lab of a 25,000-account model-theft attack, reviving calls for coordinated security policies.

1 of 5

Anthropic Accuses Alibaba

Anthropic sent a June 10, 2026 letter to Tim Scott (R-SC) and Elizabeth Warren (D-Mass.) accusing Alibaba of a distillation attack on Claude.
Alibaba’s Qwen AI lab used 25,000 fake accounts to generate 28.8 million Claude interactions between April 22 and June 5, 2026, targeting software-engineering, agentic-reasoning and long-horizon capabilities.
Anthropic called the campaign illicit, largest distillation attack, urged export controls, noted Alibaba’s stock fell ~3 percent and the Commerce Department restricted Mythos and Fable models.

1 / 5