Anthropic signs out some Claude users, removes saved payment methods, and issues refunds after infostealer malware on user PCs hijacked sessions to drain usage
Anthropic is warning some Claude users that infostealer malware on their PCs has stolen active Claude login sessions …
Context & Ripple Effects
Anthropic had already tightened access for certain Claude capabilities through identity verification requirements, while its 2025 threat report documented Claude’s use in sophisticated cybercrime. The August incident shifts the security focus from how the model is used to whether a customer’s authenticated account can be commandeered.
The response—ending affected sessions, deleting stored cards, and refunding unauthorized use—treats a stolen Claude session as both an account-security and payment-risk event.
First-order effects
- Affected Claude users lose active access and stored payment methods while Anthropic reverses charges tied to unauthorized usage.
- Anthropic contains further subscription drainage by invalidating compromised sessions and removing the payment credentials available to an attacker.
Second-order effects
- Affected users must remediate the malware on their PCs before replacement sessions or newly stored payment methods can be trusted, making endpoint security a prerequisite for uninterrupted Claude access.
- Claude’s support and billing operations must handle account recovery and disputed usage together, rather than treating session theft as a login problem alone.
Third-order effects
- If session hijacking persists, AI subscription services will need to treat authenticated browser sessions as part of the payment-security perimeter, with stronger recovery and re-authentication controls around high-usage accounts.
The trend: Consumer AI services are becoming a target for commodity credential theft because a live session can unlock both model access and metered spending.