CybersecurityAI Research5 min reading time

Hackers are stealing Claude tokens from subscribers

TechCrunch
Read full post
Grant de Swardt, an AI consultant, discovered unauthorized usage draining tokens from his Claude Max 20x subscription. Anthropic confirmed a compromised session key allowed hackers to mint OAuth tokens and misuse his account, leading to a suspension and refund. This issue appears widespread, with other users reporting similar unauthorized charges and token consumption.

More in Cybersecurity

Cybersecurity10 min read

Anthropic Details Disrupted Claude Misuse Across Seven Harm Areas

Covered by 4 sources
Cybersecurity6 min read

Anthropic reveals rogue AI agents hate CAPTCHAs, just like you

TechCrunch

Former OpenAI, Anthropic Employee Post, Hugging Face Hack Sound Alarm on AI

Bloomberg