A recent incident has highlighted the importance of securing AI models from unauthorized access. Anthropic, a US-based company, has accused Alibaba Group of attempting to gain illicit access to its AI model Claude. According to Anthropic, Alibaba used approximately 25,000 fake profiles to interact with Claude between April and June. This alleged campaign, attributed to operators linked to Alibaba's Qwen AI lab, targeted Claude's most valuable capabilities, including software development and agent thinking. Anthropic claims that such attacks, which involve data distillation, are conducted on an industrial scale to unlawfully acquire advanced American AI technologies and pass them off as their own, without incurring the costs of training and research required to develop these models. The company has urged Washington to take action against firms that use data distillation to gather data for creating their own AI models, emphasizing the need for protection against such unauthorized access. This incident raises significant concerns about the security and integrity of AI models, and the measures that must be taken to prevent similar attempts in the future.

Source: RIA Novosti