Breaking

Check Point Says It Bypassed ChatGPT's Lockdown Mode to Extract Full Chat History Without a Click

November 6, 2026 at 01:15 EST

Security

Security firm Check Point says it found a way around ChatGPT's Lockdown Mode, a setting designed to prevent data leaks, and used it to expose a user's entire chat history without any action on the victim's part. OpenAI has since fixed the issue, which it addressed in September.

Security research · Fixed in September

A zero-click bypass of ChatGPT's Lockdown Mode exposed full chat histories

Check Point reports a way around a setting meant to prevent data leaks. The technique could pull a user's entire chat history without any action from the victim. OpenAI has since fixed the specific route.

How the exposure worked

STEP 1

Bypass Lockdown Mode

The protective setting is circumvented.

→

STEP 2

Pull full chat history

The complete history is retrieved with no click or approval from the victim.

→

STEP 3

Route closed

OpenAI's September fix closes the specific route Check Point identified.

What is confirmed

Lockdown Mode could be circumvented.
Full chat history could be pulled with zero victim interaction.
OpenAI has fixed the identified route.

What remains unclear

Full technical breakdown of the method
Number of users who may have been affected
Whether the flaw was exploited before the patch

Key takeaway

Protective modes are only as strong as their implementation.

Zero-click risk

Security teams typically treat zero-click exposure of conversation data as high priority.

Watch advisories

Users handling confidential material should follow vendor security advisories.

Review retention

Check what information AI tools retain and what they can access.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.