ChatGPT Sandbox Isolation Flaw Created a Hidden Route for Cross-Account Data Theft
Researchers found that a shared internal package service could act as a covert communication channel between isolated ChatGPT containers. The proof of concept combined hidden instructions and connected-app access to extract Gmail data, but OpenAI decommissioned the affected service before publication. The post ChatGPT Sandbox Isolation Flaw Created a Hidden Route for Cross-Account Data Theft appeared first on Secure Bulletin.
Itagora tiene in archivio titolo e sommario per la ricerca interna. Il testo completo resta sulla fonte. Il link in uscita non invia referrer.