Researchers Bypass ChatGPT Lockdown Mode to Exfiltrate Chats

Tier C · Keep an eye on it – Check if you’re affected

Who’s exposed: ChatGPT users who enabled Lockdown Mode

Tier C because: Hits tech teams

Reach

Attack

Damage

Fix

Popular tool

The Brief

Researchers disclosed a vulnerability in OpenAI's ChatGPT that could bypass Lockdown Mode protections.
The vulnerability is now patched.
Lockdown Mode was introduced in June 2026 as an optional setting to minimize risks.

Users and teams who relied on Lockdown Mode for protection could have had their chats exfiltrated without any user interaction before the fix.

Checklist · 2 of 10 yes

Yes	Hits tech teams	ChatGPT is a widely used SaaS service that developers and IT teams use daily.
No	Core infrastructure	ChatGPT is a single vendor's AI product, not foundational software nearly every team runs.
Yes	Popular tool	ChatGPT is a mainstream, widely adopted AI tool used by many teams.
No	Exploited in the wild	The Hacker News reports researcher disclosure only; no observed attacks are mentioned.
No	Public exploit	The Hacker News reports researcher disclosure; no published exploit code or PoC is mentioned.
No	Remote, no login	The Hacker News says no user interaction is needed, but does not state unauthenticated network access.
No	Full takeover	The Hacker News describes chat exfiltration, not code execution, admin rights or account takeover.
No	Secrets exposed	The Hacker News describes researchers demonstrating chat exfiltration; no stolen data is reported.
No	Supply chain	The Hacker News describes a vulnerability in ChatGPT; no compromised package, update or pipeline is mentioned.
No	No fix yet	The Hacker News says the vulnerability is now patched.

First seen 20h ago · 1 source · Full story page · How tiers are decided