Skip to content
NewA calmer, faster ZeroHour
StoryVulnerabilityhigh

Shared ChatGPT Links Enable Account Takeover

2 sources · 2 articlesFirst reported Updated ()1
What's new

First merged summary.

Merged summary · x-ai/grok-4.7 · rewritten as coverage arrivesSummaries by Vani AI

Researchers showed shared ChatGPT links could take over accounts via HTML preview exploits but OpenAI fixed it on October 26.

Researchers demonstrated that a public share link to ChatGPT executes attacker-controlled JavaScript in the victim's signed-in chatgpt.com session. The attack chain exploits HTML previews using postMessage-triggered MathLive virtual keyboard, turning untrusted content into HTML in the main page via an unsandboxed srcdoc iframe and YouTube JSONP callback that bypasses CSP. This grants access to list conversations, read unshared private chats, and call authenticated APIs as the victim. OpenAI fixed the issue on October 26 and rated the report P1.

Key facts
  • Public ChatGPT share runs attacker JS in signed-in session
  • Exploits MathLive graph listener and postMessage for HTML injection
  • Bypasses CSP using unsandboxed iframe and YouTube JSONP
  • Allows reading private chats and authenticated API calls
  • OpenAI fixed on October 26, rated P1
  • Confirmed by researchers and Parameter Labs
Analyze this story with Vani AIImpact, who is affected and what defenders should do, in a new chat

Timeline

dated events in the reporting · oldest first
  1. Oct 26, 2026OpenAI fixed the issue · Hacker News

What each outlet adds

beyond the shared facts
  • Hacker NewsConfirmed listing conversations, reading unshared private chats, and API calls; write-up states fixed October 26 as P1
  • Simon WillisonDiscovered by Parameter Labs; targets interactive custom HTML apps in sandboxed iframes; previously called this Canvas
Vendors
OpenAI
Products
ChatGPTMathLive
Threat actors
Parameter Labs

Coverage

oldest first · one row per article · UTC
  1. · 1d ago Hacker News · securityFirst report

    ChatGPT share links executed JavaScript in signed-in sessions

    Researchers read a private chat and sent it off-site; OpenAI fixed the P1 issue on October 26.

  2. · 13h ago Simon Willison

    Parameter Labs finds exploit to take over ChatGPT accounts

    Parameter Labs discovered a complex attack chain exploiting ChatGPT's HTML previews via postMessage-triggered MathLive virtual keyboard.