Poison Claude Sells Discounted Claude Access While Its Operator Sees Every Customer Prompt

Cybersecurity researchers have found multiple services advertising illegal access to artificial intelligence models on underground cybercrime forums and messaging platforms. One of these services, Poison Claude, offers discounted access to certain large language models, including several versions of Opus and Sonnet. The operator of Poison Claude appears to have visibility into every customer interaction. These services are being advertised on various underground platforms. Poison Claude specifically claims to provide access to models from Anthropic. The discovery of these services suggests a demand for unauthorized access to AI models.