Hackers Steal Claude Sessions to Hijack Paid AI Accounts

Anthropic’s Claude is widely used by consumers and businesses, with paid accounts providing access to larger allocations of AI computing capacity. The attacks do not depend on cracking passwords. Instead, information-stealing malware extracts browser cookies and authenticated sessions, allowing criminals to impersonate victims and potentially bypass both passwords and two-factor authentication. That makes session credentials a valuable target wherever access to costly AI services can be resold or abused.

Anthropic recently warned that several malware families are targeting credentials tied to Claude accounts. Users may first notice that their usage allowance has been depleted unusually quickly, even though an attacker has already taken control and consumed paid AI capacity. The company has not disclosed the number of affected accounts, the financial losses involved or a precise date for the warning. Users are advised to terminate suspicious sessions, sign in again and inspect their devices for malware.