AnalysisCybersecurityNovember 5, 2026

Check Point bypassed ChatGPT Lockdown Mode to exfiltrate chats

Read original source →thehackernews.com

Check Point found a sandbox-escape in ChatGPT's HTML code-block previews: nested frames reached an unmodified popup function, letting content open outside the sandbox and evade Lockdown Mode's outbound-request limits. The flaw allowed zero-click chat data exfiltration in the desktop app; OpenAI has patched it.

1 source

More stories today

Open the live feed