Anthropic says Alibaba must be punished for largest Claude cloning attack

Key Points:

• Anthropic has accused Chinese firm Alibaba of conducting the largest known attack to illicitly extract capabilities from its AI model Claude, generating nearly 29 million exchanges via 25,000 fraudulent accounts between April and June.
• The campaign allegedly targeted Claude’s advanced functions like agentic reasoning and software engineering, using obfuscation and proxy networks to evade detection, as China intensifies efforts to match US AI capabilities following Anthropic’s Mythos release.
• Despite US government warnings and export controls initiated under the Trump administration, Alibaba—listed on the NYSE and accountable to US investors—continued these attacks, prompting Anthropic to call for stronger legislative measures to deter such illicit distillation.
• Alibaba has denied military ties and filed a lawsuit against the Trump administration’s blacklist designation, while Anthropic warns that without intervention, these attacks could accelerate China’s AI advancements and pose national security risks.
• Chinese tech leaders acknowledge the strategic threat posed by US frontier AI models like Mythos, with calls to develop comparable domestic systems to ensure cyber defense parity, highlighting the escalating AI rivalry between the US and China.