AI agents vulnerable to prompt injection via emails, websites; Gen Digital unveils Agent Trust Hub

Security firm Gen Digital highlights that AI agents can be hijacked through malicious emails, websites, and plugins, abusing their legitimate permissions to execute commands, steal data, or delete files. The company introduces Agent Trust Hub with skill verification, secure connectivity, and runtime controls to mitigate such attacks. This underscores a growing attack surface as enterprises deploy autonomous AI agents.

AI disclosure: Model-written summary of the source, not a verified fact. Check the sources. How Mosaic works