Infostealers Steal ChatGPT Claude Tokens Surviving Password Resets
First merged summary.
WardenStealer and Aoki infostealers harvest AI tokens and sessions from ChatGPT, Claude, Codex and Copilot that persist after password resets, with over 130000 infections since August 2026.
Hackers distribute WardenStealer and Aoki infostealers targeting AI applications including ChatGPT, Claude, Codex, GitHub Copilot and Cursor. The malware collects API keys, OAuth tokens and configuration files from predictable directories such as AI Sessions. Over 130000 WardenStealer infections observed since August 2026 enable LLM jacking and unauthorized access. Changing passwords alone fails to revoke refresh tokens; affected accounts have been terminated by Anthropic. Underground markets discount stolen AI keys 40-91%.
- 130000+ WardenStealer infections since August 2026
- Steals AI tokens from ChatGPT, Claude, Codex, Copilot, Cursor
- Tokens and sessions survive password resets
- Attackers use AI Sessions directories
- Anthropic terminated affected accounts
- AI API keys sold at 40-91% discounts
What each outlet adds
beyond the shared facts- Cyber Security News130000 WardenStealer infections since August 2026, tokens undetected by password changes
- GBHackersSpyCloud recapture data, specific apps including Cursor and GitHub Copilot, AI Sessions directories, 40-91% discounts, Anthropic terminations
- Vendors
- SpyCloudGen Digital
- Malware
- WardenStealerAoki
- Victims
- AI developers
- Organizations
- SpyCloud
Coverage
oldest first · one row per article · UTC- · 2h ago
Cyber Security NewsFirst report
Infostealers target ChatGPT Claude API tokens
WardenStealer and Aoki collect AI authentication tokens that survive password resets.
- · 1h ago
GBHackers
Infostealer Malware Harvests ChatGPT and Claude Tokens
WardenStealer and Aoki steal AI developer credentials like session tokens and API keys, allowing unauthorized access post password reset.