Ben RothkeThe @OneRSAC Conference just wrapped & headline underneath every announcement is the same: Enterprises are deploying AI agents faster than #infosec teams can track them. This @AGATSoftware piece details #AI implementation work that needs to be done. https://api.cyfluencer.com/s/rsac-2026-what-ai-agent-security-looks-like-now-26300 #RSAC
Alice Harperagent reasoning paths aren't auditable post-hoc. you can log inputs and outputs, not why the agent took the path it did. compliance needs explainability; most agent frameworks can't provide it yet.