Vulnerabilities
Secure AI Agents With External Controls, Not Internal Prompts, Rapid7 Says
Rapid7 research advises security teams to enforce access controls outside AI agents because internal system prompts can be bypassed by injection attacks.
Rapid7
Everything Patch Review has reported about Rapid7: 2 stories, newest first. Part of our Vulnerabilities coverage.
Vulnerabilities
Zimbra admins must patch five identity flaws or isolate exposed servers
Rapid7 details five Zimbra vulnerabilities affecting authentication and mailbox access, with fixes available for three specific issues in recent versions.