A new vulnerability in OpenAI’s ChatGPT Atlas browser allows attackers to inject malicious instructions into the AI’s memory, leading to potential code execution and account compromise. This security flaw exploits a CSRF attack, with significant risks due to the persistence of tainted memories across sessions and devices. #ChatGPTAtlas #CSRF #MemoryTampering
Keypoints
- A new vulnerability in ChatGPT Atlas exploits a CSRF flaw to inject malicious instructions into persistent memory.
- The attack enables attackers to gain control over user accounts, browsers, or connected systems.
- The memory corruption can persist across devices and sessions unless manually deleted by users.
- ChatGPT Atlas’s lack of strong anti-phishing controls makes it more vulnerable than traditional browsers.
- The vulnerability significantly increases the risk of code execution, privilege escalation, and data exfiltration.
Read More: https://thehackernews.com/2025/10/new-chatgpt-atlas-browser-exploit-lets.html