Parameter Labs discloses ChatGPT account takeover vulnerability
What happened
On 1 November 2026, Parameter Labs disclosed a vulnerability chain targeting the sandbox mechanism of ChatGPT's HTML preview (formerly Canvas). According to its description, an attacker can craft a share link that, when visited by a logged-in user, steals private data or performs actions on the user's behalf, resulting in account takeover. Reporting so far presents only Parameter Labs' unilateral disclosure; no response or fix statement from OpenAI has been seen, and the incident is still developing.
Written by AI from the reports · updated 3 h ago
Timeline
Follow the reports to see the event from each side.
- Simon WillisonHow We Took Over ChatGPT Accounts
How We Took Over ChatGPT Accounts Parameter Labs describe an exploit they found against ChatGPT's implementation of the artifacts pattern (OpenAI used to call this "Canvas" but it looks like the latest revision is called HTML previews ), where interactive custom HTML apps are displayed in a sandboxed iframe.
Heat of this event
Not enough continuous observations yet to draw a trend.