Overview A newly uncovered vulnerability in OpenAI’s ChatGPT Atlas Browser Mode has raised significant concern within the global cybersecurity community. Security researchers discovered that attackers can exploit this feature to inject persistent hidden commands directly into the AI’s operational memory. This exploit, if leveraged, allows threat actors to execute unauthorised actions, retrieve sensitive data, or...