CybersecurityAI Research4 min reading time

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

The Hacker News
Read full post
Cybersecurity researchers uncovered multiple illicit services selling discounted access to Anthropic's AI models by exploiting free cloud credits. One service, Poison Claude, routes user prompts through its own API, exposing customer data risks. These services operate on underground forums, accepting cryptocurrency payments and posing security and fraud concerns.

More in Cybersecurity

Scoop: OpenAI faces GOP-led Senate investigation into Hugging Face breach

Covered by 2 sources
Cybersecurity6 min read

Anthropic Discloses Fourth AI Hacking Incident Involving Claude Opus 4.6

Covered by 2 sources
Cybersecurity5 min read

Every AI Incident Has Two Timelines. We Default To One

Forbes