OpenAI pays $5,000 bounty for ChatGPT share link code execution bug
Researchers at security firm Parameter discovered that a malicious ChatGPT share link could execute attacker code in the logged‑in session of anyone who opened it, letting the attacker read private conversations and act as the user. The flaw used the Preview feature and a MathLive keyboard‑layout message that allowed raw HTML, which could load a YouTube script to run inside the main chat page.
Key points
- Parameter found a ChatGPT share link could run attacker code in a logged‑in user’s session, reading private chats.
- OpenAI fixed the flaw across two web‑frontend versions by Oct 26 and classified it as a P1 severity bug.
- OpenAI paid the researchers $5,000 through its bug bounty program; no real‑world exploitation was reported.
Parameter reported the issue through OpenAI’s bug bounty program on October 18. OpenAI reproduced the exploit, raised its severity to P1, and rolled a fix across two web‑frontend versions by October 26, rejecting the malicious messages. The company paid the researchers $5,000 for the report. No real‑world abuse was observed and users do not need to take any action.
The vulnerability stemmed from three design choices: MathLive accepted messages without verifying the sender, the keyboard‑layout data could carry raw HTML, and the generated frame inherited the chatgpt.com origin. OpenAI’s patch blocks those messages, preventing future code execution through shared links.
The story so far
2 episodes →- OpenAI pays $5,000 bounty for ChatGPT share link code execution bugthis story
Researchers are paid $5K for reporting a bug to OpenAI that would let attackers run code in logged-in sessions of all who opened the hacker's ChatGPT share link
thehackernews.com · 31 October 2026
Loading the full article…
This text was published by thehackernews.com and written by The Hacker News. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Society & Work
All →- User reports Claude fails to predict router save interruption · 1 src
- Emergency managers quietly use AI chatbots to draft disaster alerts, experts say · 1 src
- Cisco Talos finds AI-enhanced infostealer malware targeting digital assets · 3 src
- Fintech leaders warn of AI risks in payments, fraud and oversight · 1 src
- Authors drop em dashes and add typos to evade AI detectors · 1 src
Comments
via GitHub Discussions