Topic starter
15/08/2026 6:09 pm
TL;DR: RSA 2026 Day 1 sessions, as covered by AppSOC, argue that agentic AI is compressing detection and attack timelines beyond human-scale response, while enterprises still rely on visibility, review, and governance models built for slower systems. The core implication is that identity and security programmes must shift from static oversight to context-aware, continuously enforced control.
Editorial analysis by NHI Mgmt Group, based on content published by AppSOC: “RSAC 2026 AI Security Insights”.
Key questions
Q: How should security teams govern AI agents without creating a manual review bottleneck?
A: Use policy, automation, and class-based controls so agents are provisioned through deployment pipelines, not ticket queues.
Q: Why do traditional detection and response models struggle with agentic AI threats?
A: They depend on human-paced observation, triage and escalation, while agentic threats can scan, decide and move in parallel.
Q: What breaks when organisations rely only on periodic access reviews?
A: Periodic reviews miss access that changes between certification windows, which leaves risk hidden until after the fact.
Practitioner guidance
- Rebuild response around machine-speed decisions Identify where detection, approval and remediation still depend on manual triage, then shift those controls to enforcement points that can act during the session rather than after it closes.
- Inventory agent and machine identities as active actors Map every autonomous agent, API and service identity to an owner, purpose, privilege scope and revocation path so hidden executors do not outlive the workflow they support.
- Replace periodic review with context-aware enforcement Use runtime context from asset relationships, data sensitivity and execution conditions to decide whether an agent can continue, rather than waiting for the next access review.
Bottom line: Agentic AI changes the security tempo so completely that human-paced detection and response become weak controls rather than dependable safeguards.
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →
15/08/2026 6:17 pm
Human-speed security is now a broken operating assumption. The article shows that detection and response programmes built around analyst review, alert triage and post-facto remediation no longer match the tempo of agentic adversaries. That is not a tooling problem alone; it is a control-model mismatch. Security teams should treat response latency as a structural risk, not a staffing inconvenience.
A few things that frame the scale:
- 67% of organisations still rely heavily on static credentials despite the risks they pose to agentic AI deployments, according to the 2026 Infrastructure Identity Survey.
- 53% of security leaders expect AI to run major portions of their infrastructure autonomously within the next three years, according to the 2026 Infrastructure Identity Survey.
A question worth separating out:
Q: What does context-aware security add for agentic AI and non-human identities?
A: It links identity, data and execution conditions so controls can follow how risk actually moves through the environment. Instead of treating each alert or account in isolation, context-aware security shows which identities, permissions and dependencies matter at that moment. That is essential when one agentic action can cascade into many downstream effects.
👉 Read our full editorial: AI-driven threats expose the limits of human-speed security