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

Cybersecurity researchers discovered more than six services advertising illegal access to artificial intelligence models on underground cybercrime forums and messaging platforms. Poison Claude claims to provide access to Anthropic models and routes customer prompts through accounts controlled by the service to Anthropic. Researchers said that, because the service operates as a gateway proxy, its provider has full visibility into prompts, creating a risk that data could be leaked or sold.