Foreign firms illegally copy Anthropic's AI tech via dark web, threatening US AI leadership.

Anthropic alleges foreign competitors, especially from China, illegally access its AI models through a process called distillation, using dark web methods and stolen accounts to train cheaper copycat AI systems. This practice involves creating thousands of fraudulent accounts to extract data, raising intellectual property and national security concerns. Anthropic is pushing for stricter controls as it prepares for a potential IPO, highlighting the broader industry challenge of protecting AI innovations from illicit use amid global competition. The US government has also flagged such practices as unacceptable and is considering measures against foreign actors.