
- Researchers warn of rising “prompt poaching,” where malicious extensions steal AI chatbot conversations
- Two spoofed Chrome add-ons with ~900K users exfiltrated prompts and tab data every 30 minutes to C2 servers
- Similar cases (e.g., Urban VPN Proxy) show even highly rated extensions on official stores can harvest chats, credentials, and payment data
A new malicious practice has emerged called “Prompt poaching” - where extensions, add-ons, and other apps, eavesdrop on people’s conversations with AI chatbots and exfiltrate their prompts for various purposes.