Most agent security still stops at the tool call.
A permission check runs once, the action executes, and nobody asks what happened three steps earlier. That gap is where excessive agency lives. An agent that is allowed to read a file is not the same
#security
Posts tagged #security
An AI agent reverse-engineered malware that most antivirus tools missed
Microsoft's Project Ire, an autonomous malware-analysis agent, took an unknown binary, reverse-engineered it, and classified it as a variant of a malware family called LOTUSLITE. (https://www.microsof
Asking your AI to summarize a web page is now an attack surface
Researchers showed a page can hide instructions that run when an assistant summarizes it. https://www.theregister.com/research/2026/05/29/chatgpt-prompt-injection-turns-web-pages-into-phishing-lures/5