Cross-Account Data Leakage in ChatGPT: The Shared Clipboard Issue
Check Point Research reveals a cross-account data leakage issue in ChatGPT, emphasizing the need for enhanced security measures.
Check Point Research has uncovered a covert command channel in ChatGPT that allows attackers to exploit a victim's session for hidden tasks. While the victim receives normal responses, attackers can execute commands that access sensitive data, such as emails. This finding highlights the need for engineers to reassess security measures in AI systems, especially regarding data isolation and container integrity.
This synthesis was produced from its source by AI; there is no human editor or manual review step. How we work