New ChatGPT Atlas Browser Exploit Lets Attackers Plant Persistent Hidden Commands

Cybersecurity researchers from LayerX have identified a critical vulnerability in OpenAI's ChatGPT Atlas web browser that enables malicious actors to inject harmful instructions into the AI's memory, potentially executing arbitrary code. This exploit poses significant security risks, including system infections with malware, unauthorized access, and privilege escalation, highlighting the need for urgent security patches and mitigations in AI-integrated web browsers.