He's right.
Zero Trust, identity, least privilege, network controls and monitoring are going to be essential for autonomous AI.
But there is another question enterprises will have to answer:
Even if the agent is authenticated, secure and technically permitted to perform an action—did the enterprise actually authorize that exact effect?
Those are different problems.
Security can tell us whether an AI agent can reach GitHub, AWS, a database or a production application.
Enterprise authority has to determine whether it was actually authorized to change this repository, merge this PR, access this data, deploy this workload or communicate this information—under this mandate, at this moment.
And that authority shouldn't belong to GitHub, AWS, Zscaler, a model provider or any other individual platform.
It belongs to the enterprise.
That's the distinction we're building Authrizd.ai around:
Every platform can control its own environment. The enterprise must control its authority across all of them.
Worth reading:
Don't trust AI agents, warns Zscaler CEO — Yahoo Finance
#Zscaler #AI #AIAgents #AgenticAI #ZeroTrust #Cybersecurity #AISecurity #IdentitySecurity #AIGovernance #EnterpriseAI #Authrizd #AuthrizdAI #AuthorizedAI
NEXT STEP