Anthropic warns of token theft targeting Claude subscribers

A user of Anthropic’s Claude AI platform reported that his account was consuming tokens in the background last month, even though he had not interacted with the system. The unexpected usage prompted the user to investigate, and he found no legitimate activity on his account that would explain the token drain.

In response, Anthropic issued an advisory to its user base warning that unauthorized token consumption could be the result of hacking attempts. The company urged users to review their account activity logs, change passwords, and enable multi‑factor authentication to mitigate potential security breaches. Anthropic also announced plans to investigate the incident and strengthen monitoring of token usage across its services.

The warning highlights the growing importance of security in AI‑driven platforms, where token usage can directly translate into costs for users. Anthropic’s prompt communication aims to reassure customers that the company is actively addressing the issue and working to prevent similar incidents in the future.