Identity security is shifting from visibility to enforcement at RSAC

At a glance

What this is: This is a Silverfort recap of RSAC identity security themes, arguing that the market is moving from visibility into identities toward inline enforcement at access time.

Why it matters: It matters because IAM, PAM, and NHI programmes now need controls that decide before execution, not just inventories that describe who and what exists.

By the numbers:

• Through 2028, at least 80% of unauthorized AI agent transactions will be caused by internal violations of enterprise policies rather than malicious attacks, according to Gartner research cited by Silverfort.

Context

Identity security is moving from discovery to enforcement. The operational question is no longer only which identities exist or what they can access, but whether access can be evaluated and constrained before the identity acts. That matters because traditional identity controls were built around slower, human-paced access patterns, not identities that can appear, act, and disappear quickly.

The same shift is affecting privileged access management, where vaulting credentials is no longer enough to explain or control how privilege is used in real time. In hybrid environments, privilege is not confined to a small tier of administrators. It now spans human, non-human, and increasingly agentic actors, which changes the governance problem from storage to enforcement.

Key questions

Q: What breaks when an identity security tool only provides visibility?

A: Visibility without lifecycle action leaves teams with better reporting but the same exposure. If the platform cannot drive offboarding, revocation, or certification workflows, it may improve awareness while privilege drift, orphaned access, and stale entitlements continue unmanaged.

Q: Why do vaults alone not solve privileged access risk?

A: Vaults protect where credentials live, but they do not control how those credentials are used once an identity is active. Privilege risk often emerges at execution time, inside hybrid and agentic workflows that never depend on the vault after checkout. Teams need controls that govern use, not only storage, if they want actual privilege reduction.

Q: How should organisations govern AI agent access without losing operational speed?

A: Use task-scoped access, explicit human ownership, and runtime monitoring together. Fast-moving agents still need a clear approval path, a defined purpose, and revocation when behaviour drifts. The goal is not to slow automation, but to make every action traceable and every entitlement reviewable before it becomes standing risk.

Q: What is the difference between discovering identities and enforcing access decisions?

A: Discovery tells you what identities exist and what they can reach. Enforcement decides whether a request should be allowed, denied, or constrained before action occurs. The distinction matters because many identity programmes stop at inventory, which improves visibility but does not reduce the chance of misuse.

Technical breakdown

Why visibility alone no longer governs identity risk

Visibility tells you what identities exist, what entitlements they hold, and where they appear in the environment. Enforcement adds the decision layer: whether a request should be allowed, blocked, or constrained at the moment of use. That distinction matters because inventory does not stop misuse. In identity-heavy environments, especially where workloads, service accounts, and AI agents act quickly, the security boundary is no longer the directory record. It is the access decision made inline, at authentication or authorisation time. Programmes that stop at discovery often know the estate but not the control outcome.

Practical implication: move from identity inventory reporting to inline access decisioning.

Why vaulting is not the same as privileged access control

Vaulting centralises secrets, but it does not necessarily govern how privilege is consumed after authentication. Once a credential is issued or retrieved, the risk shifts to the session, the request path, and the action taken by the identity. That is why vault-centric PAM can miss the real control point in environments where privilege is ephemeral, federated, or machine-driven. The article’s argument is that privileged access must be enforced in real time across the authentication flow, not only protected at rest in a vault. This is especially relevant when privilege appears outside traditional Tier-0 assumptions.

Practical implication: treat secret storage as necessary but insufficient for privileged access governance.

How AI agent identity changes the enforcement model

AI agents are not just another automation layer. When they can make decisions and execute tasks, they behave as runtime identities whose risk depends on policy, context, and allowed action scope. That changes how organisations should think about access governance, because an agent can trigger transactions faster than a human reviewer can intervene. The quoted Gartner figure in the article reinforces that policy violations, not just external attackers, will drive much of the risk. Identity security therefore has to incorporate policy enforcement for AI agents as an access-control problem, not only as an AI governance problem.

Practical implication: define agent usage policy and enforce it at the point of execution.

Breaches seen in the wild

• Scania insurance portal breach 2025: An attacker used an external user login, likely stolen by infostealer malware, to take insurance claim documents from a Scania portal.

Read our 52 NHI Breaches Analysis report for a comprehensive view of breaches impacting Non-Human Identities including AI Agents.

NHI Mgmt Group analysis

Identity governance is moving from observability to control enforcement. The old model assumed that seeing identities and their entitlements was the hard part. In practice, visibility without an inline decision point leaves organisations with a map of risk, not a way to stop it. The practitioner conclusion is that identity programmes now have to prove control at the moment access is requested, not only after the estate has been catalogued.

Vault-centric PAM reflects a control assumption that no longer matches how privilege is used. Vaults are useful for secret custody, but they do not solve the governance question of how privilege behaves during live access. That assumption works only when access is slow, bounded, and human-paced. In hybrid environments with machine and agentic identities, privilege is dynamic and distributed, so enforcement has to follow the transaction path, not just the credential store. The practitioner conclusion is to re-evaluate where PAM actually exerts control.

AI agent security is becoming an identity problem, not just an AI problem. The article’s strongest signal is that agent behaviour is now part of access governance, because agents can initiate actions that outpace human review. That collapses the old separation between AI oversight and identity enforcement. The practitioner conclusion is that teams need a governance model that binds acceptable use, access scope, and policy enforcement together.

Privilege is no longer a Tier-0 concept, which means selective coverage leaves gaps. The article points to a world where privileged access exists across hybrid environments and across more identity types than traditional admin accounts. That broadens the attack surface and makes narrow coverage strategies brittle. The practitioner conclusion is that identity control must be consistent across human, non-human, and agentic actors, not reserved for legacy admin pathways.

Real-time enforcement will define the next phase of identity security architecture. The market signal is not simply that new tools are emerging, but that enforcement is becoming the architectural requirement. That aligns with NIST CSF 2.0 access control outcomes and with NHI governance patterns that emphasise access permissions and entitlement decisions. The practitioner conclusion is to judge identity controls by whether they can constrain action at runtime.

What this signals

Inline enforcement is becoming the practical test for identity maturity. Programmes that can enumerate identities but cannot block risky access are only partially governed. The next maturity step is proving that policy can shape behaviour at the moment of authentication or transaction initiation.

AI agent access will force identity teams to unify policy, access, and governance. Once agents can act independently, acceptable-use rules become access-control rules. That pushes identity teams to coordinate human policy education, technical enforcement, and runtime control in one programme.

Selective coverage is the wrong response to hybrid privilege. Privilege now appears across service accounts, workloads, and agentic actors, so controls that only cover legacy admin pathways will miss a growing share of risk. Identity teams should assume privilege is distributed until proven otherwise.

For practitioners

• Define enforceable AI agent usage policy Write explicit rules for what AI agents may access, what actions they may trigger, and which transactions require blocking or step-up control. Make the policy operational enough to drive enforcement logic, not just acceptable-use language.
• Move PAM control points inline Evaluate whether privileged access decisions are being made at authentication and session start, rather than only at secret retrieval. Where vaults exist, treat them as one control layer, not the control model itself.
• Review privilege coverage across hybrid identities Map where privileged access exists outside legacy admin accounts, including service accounts, workload identities, and AI-driven actors. Look for coverage gaps created by selective Tier-0-only controls.
• Separate visibility from enforcement metrics Track whether identity controls actually block or constrain action, instead of measuring only inventory completeness, entitlement count, or discovery coverage. Use control outcome metrics to prove governance.

Key takeaways

• The core identity-security problem is no longer visibility, but whether the programme can stop or constrain access before action.
• Privilege now extends well beyond human administrators, which makes vault-only approaches too narrow for hybrid environments.
• AI agents turn policy into a runtime control issue, so acceptable use and enforcement have to be linked in the operating model.

Key terms

• Inline Enforcement: Inline enforcement is the technical act of applying access policy in the live session path, not just at approval time. It matters because identity governance without runtime enforcement can authorize access that the session layer never actually constrains, especially in distributed and third-party environments.
• Vaultless PAM: Vaultless PAM is a privileged access model that reduces reliance on credential vaulting as the main control point. Instead of protecting secrets and hoping that checkout use is safe, it enforces access decisions inline at authentication and authorization time. That makes it more aligned to hybrid and fast-moving identity behaviour.
• Agentic Identity Governance: The discipline of managing, governing, and auditing the identities of autonomous AI agents across their full lifecycle, from provisioning with least-privilege credentials through continuous monitoring and decommissioning. An emerging sub-discipline of NHI governance.
• Identity Enforcement Gap: The identity enforcement gap is the space between knowing an identity exists and being able to stop it from acting. It appears when inventory, entitlement visibility, or vault custody are treated as substitutes for runtime control, leaving misuse technically visible but operationally allowed.

Deepen your knowledge

NHI governance, agentic AI identity, and machine identity lifecycle are core topics in our NHI Foundation Level course, the industry's only accredited NHI security programme. If you are building or maturing an IAM programme, it is worth exploring.