Attacks exploiting the Plex2Shell chain are live in the wild, installing cryptominers and residential proxy software on unpatched…
Researchers found a macOS ChatGPT flaw that let attackers take over the app and read every chat log.
A security firm says simply picking a model triggered arbitrary code, with no CVE published.
A fix written by GitHub's AI autofix bot left Snowflake's CI pipeline open to command injection, Wiz Research…
Mindgard researchers discovered that ChatGPT can be tricked into creating graphic sexualized and violent images using a simple…