tech
Anthropic says Alibaba must be punished for largest Claude cloning attack
Alibaba allegedly used 25,000 accounts to mine Claude over 28.8 million exchanges.
TL;DR
- Anthropic claims Alibaba launched the largest attack to clone its AI model, Claude, using over 28.8 million exchanges from nearly 25,000 fraudulent accounts.
- The alleged attacks, occurring between April and June, targeted Claude's advanced capabilities like agentic reasoning and software engineering.
- Anthropic states Alibaba used obfuscation techniques and proxy networks to evade detection, contributing to a "growing circumvention economy."
- The company argues that such attacks turn US AI investments into a subsidy for geopolitical competitors.
- Anthropic is calling for legislative changes, including updated antitrust laws, export controls on chips, and penalties for Chinese labs to deter illicit distillation attacks.
- The accusations come amid China's race to match US AI capabilities, with Chinese tech founders likening advanced models like Anthropic's Mythos to "cyber nuclear weapons."