Researchers Bypass ChatGPT Lockdown Mode to Exfiltrate Chats
Original title: Researchers Bypass ChatGPT Lockdown Mode to Exfiltrate Chats Without User Interaction
Bypass uses nested frames in HTML previews to escape sandbox and enable silent data theft; patched September 18 2026.
Check Point researchers disclosed a vulnerability in OpenAI ChatGPT that bypasses Lockdown Mode protections introduced in June 2026. The flaw allows HTML content rendered in code blocks to escape sandbox restrictions via popup windows, enabling exfiltration of chats without user interaction. OpenAI patched the issue on September 18 2026. The issue affects the desktop application and could facilitate phishing or malware delivery.
- Nested sandboxed frames in HTML previews enable escape of Lockdown Mode.
- Silent chat data theft possible via crafted prompts or shared chats.
- Vulnerability not limited to desktop environment.
- OpenAI addressed the issue on September 18 2026.
Impact: exfiltrate chat data or deliver malware via popups
From the original article
7 min read at thehackernews.com
Cybersecurity researchers have disclosed details of a now-patched vulnerability that could be exploited to bypass Lockdown Mode protections in OpenAI ChatGPT. Lockdown Mode was introduced by the artificial intelligence (AI) company in June 2026 as an optional setting to minimize data exfiltration risks arising from prompt injections. "The flaw could enable silent chat data theft, phishing,
thehackernews.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.