AIToday

Anthropic accuses Alibaba of the largest Claude cloning attack yet, using nearly 25,000 fraudulent accounts to extract the model's capabilities over two months.

Ars Technica AI8h ago5 min read
Anthropic accuses Alibaba of the largest Claude cloning attack yet, using nearly 25,000 fraudulent accounts to extract the model's capabilities over two months.

Key takeaway

Anthropic has disclosed that Alibaba conducted the largest campaign to extract Claude's capabilities, generating over 28.8 million exchanges through nearly 25,000 fraudulent accounts between April and June. The attack occurred after Trump's warnings against AI model cloning and highlights what Anthropic views as a systematic effort by Chinese companies to avoid the costs of training frontier models independently. Anthropic is calling on Congress to strengthen export controls, update antitrust law, and penalize such behavior to protect US AI leadership.

Summaries like this, in your inbox every morning.

Sign up free →

3 Key Points

  • What happened

    Between April 22 and June 5, operators affiliated with Alibaba and Alibaba Qwen generated more than 28.8 million exchanges with Claude through almost 25,000 fraudulent accounts, violating Claude's terms of service. The campaign targeted capabilities such as agentic reasoning, software engineering, and long-horizon tasks, using obfuscation techniques and proxy networks to evade detection.

  • Why it matters

    Alibaba's alleged attack comes weeks after Donald Trump warned that cloning US frontier AI models was unacceptable and a threat to national security. Anthropic argues that these distillation campaigns turn billions of dollars in American investment into a subsidy for geopolitical competitors and could accelerate China's access to advanced AI capabilities without the training costs.

  • What to watch

    Anthropic has recommended three policy steps to Congress: updating antitrust laws to allow AI firms to share information about Chinese tactics, increasing export controls on advanced chips, and passing laws penalizing Chinese labs' reliance on distillation attacks. The company warns that without intervention, these attacks could help China reach Mythos Preview-level capabilities sooner.

FAQ

How did Alibaba hide the cloning attempt from detection?
Alibaba allegedly used obfuscation techniques and proxy networks to evade detection during the campaign.
What does Anthropic want the US government to do in response?
Anthropic recommended that Congress update antitrust laws to allow AI firms to share information about Chinese tactics, increase export controls on advanced chips, and pass laws penalizing Chinese labs for relying on distillation attacks—such as limiting their access to US models, advanced chips, or data centers outside China.
When did this cloning attack take place?
The attacks occurred between April 22 and June 5.

Discussion

No comments yet. Be the first to share your thoughts!

Log in to join the discussion

Related Articles

Stay ahead with AI news

Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.

Get Started Free

Free · takes 30 seconds · unsubscribe anytime

5 minutes a day. The AI essentials.

200+ sources · Email / LINE / Slack

Get it free →