Story perspectives

Anthropic Accuses Alibaba of Fake-Account AI Attack; Shares Slide 4.4%

6/26/2026

Summary

Anthropic Accuses Alibaba

Anthropic accused Alibaba of a distillation attack on Claude using about 25,000 fake accounts for 28.8 million interactions from April 22 to June 5 2026.
Anthropic mailed a June 10 letter to Senators Tim Scott (R-SC) and Elizabeth Warren (D-Mass.) detailing the claims.
The alleged attack focused on Claude’s software-engineering, agentic-reasoning and long-horizon task abilities.
Alibaba shares dropped 4.4% in Hong Kong, underperforming the Hang Seng Tech Index’s 1.6% fall, with the firm offering no comment.

Summary

Alibaba Sues Pentagon

Alibaba sued the U.S. government in June 2026 to leave the Pentagon’s 1260H blacklist.
The Defense Department will bar purchases from 1260H-listed firms after 30 June 2026, endangering Alibaba’s contracts.
China retaliated the same day by sanctioning ten U.S. companies tied to its military.
Britain asked for a waiver on the U.S. ban of Anthropic’s AI models, whose Mythos model was flagged because it can exploit flaws.

Summary

Anthropic Halts Models

Anthropic disabled its Mythos and Fable AI models after the Commerce Dept imposed restrictions on June 12.
Anthropic announced a partnership with AI firms to counter distillation attacks.
Labs DeepSeek, Moonshot AI and MiniMax generated 150,000, 3.4 million and 13 million Claude exchanges.
KELA warned in 2025 that Alibaba’s Qwen2.5-VL model was vulnerable to prompt-injection exploits, while policy head Sarah Heck said the April-22-June-5 operation could speed Chinese models toward Mythos Preview capabilities.

History

Historical Context

In May 2019 the U.S. Department of Commerce added Huawei and 70 foreign subsidiaries to the Entity List, cutting off American chips and software. The move sparked a series of export controls, culminating in the 2025 inclusion of Alibaba on the Section 1260H blacklist. Officials now warn that similar rules could soon target AI models, echoing the Huawei precedent.

1 of 4

Anthropic Accuses Alibaba

Anthropic accused Alibaba of a distillation attack on Claude using about 25,000 fake accounts for 28.8 million interactions from April 22 to June 5 2026.
Anthropic mailed a June 10 letter to Senators Tim Scott (R-SC) and Elizabeth Warren (D-Mass.) detailing the claims.
The alleged attack focused on Claude’s software-engineering, agentic-reasoning and long-horizon task abilities.
Alibaba shares dropped 4.4% in Hong Kong, underperforming the Hang Seng Tech Index’s 1.6% fall, with the firm offering no comment.

1 / 4