What breaks when zero trust meets autonomous agents?
Your zero trust program assumes a human asks for access. Agents now request, pay, and act on their own, faster than your security team can watch.
Built for people, hit by software
The next phase of zero trust cannot focus only on proving that an identity is legitimate. Zero trust runs on a simple premise: never trust, always verify. Writing in Computer Weekly, Gartner's Niyati Daftary notes the model was built for a world where a human usually requests access. She argues most zero-trust programs have not kept pace with agent and LLM risks. Forbes Councils contributor Vikram Pande agrees. He argues the next phase must decide whether a human or machine identity should perform a specific action on a specific resource, under current conditions.
- Verification at login is not enough
- Authorization must be per action, per context
Agents act like neither users nor services
Agents act in ways that were never considered as part of zero trust principles. Teleport's Chris Webber told Help Net Security that agents act fast like software, unpredictable like people, and continuously pursue goals. Point-in-time verification and static least privilege fall short for that behavior. Security teams must stop spotting anomalies after the fact. Instead, they should enforce rules at every step an agent takes, starting agents with zero privileges.
- Agents iterate relentlessly toward a goal
- Static permissions go stale in minutes
Agents need onboarding, not just credentials
None of that exists for agents. Per VentureBeat, every agent needs a verifiable identity, a credential it cannot leak, and narrow permissions before touching production. Most IAM platforms were not built to issue any of those to software. The article points to a missing equivalent of human onboarding. Companies hire people over weeks with interviews and background checks, but deploy agents in minutes.
- Agents already outnumber planned human identities at many firms
- Trust needs a process, not just a credential
Money exposes the accountability gap
The law, however, still assumes a human approves every transfer. Yahoo Finance reports that legal experts on BeInCrypto's council see agents already paying for data, API calls, and compute on their own. Current law still assumes a human approves each transfer, they argue. When a machine pays, nobody is sure who is liable. That gap now sits at the center of the agent identity debate, they say.
- Payments law predates autonomous software
- Liability for agent transactions is undefined
What this does not solve
Can zero trust coexist with autonomous agents in typical enterprise environments? Technically, yes. In practice, not likely. CSOonline argues zero trust and autonomous agents may not coexist in practice, given board pressure to ship agentic ROI fast. The article also notes unresolved accountability when agents go rogue. None of these sources offer a finished framework. Governance, liability, and behavioral enforcement remain open problems. Vendor proposals like trusted runtimes are early.
- Board urgency outruns security maturity
- Accountability for rogue agents is still unclear
Who in your organization could today tell you how many non-human identities touch production systems?
If an agent made an unauthorized payment tonight, who would be accountable under your current policies?
Inventory every non-human identity with access to a production system in the next ten minutes. Count them.
Sources
Recommended for you
- Security
Why do API keys keep leaking from automation platforms?
Your automation tools need secrets to work, so keys get pasted into code, configs, and CI wherever they run fastest. Central storage and rotation fix it, but only if you build them.
- Security
Why does your security review still arrive too late to help?
Your feature dies in review because standards like OWASP Top 10 and ASVS were treated as final gates, not design inputs.
- Security
Why does your security checklist still miss your biggest threats?
Checklists patch known holes. Threat modeling finds the holes you never listed, by asking four simple questions about your system.
- Security
Why does AI agent access still leave your secrets exposed?
Agents need credentials to act, but agents holding secrets is the new attack surface. Proxy-brokering access keeps keys out of reach.
Enjoyed this article?
Subscribe for new articles. No spam. Unsubscribe anytime.