CrowdStrike has launched Continuous Identity for AI Agents, a security capability designed to replace standing privileges with real-time, risk-based authorisation as AI agents take on more autonomous work inside enterprises.
CrowdStrike said the new Falcon Next-Gen Identity Security capability continuously evaluates who owns an AI agent, who is calling it and the risk posture of the calling device before granting access, rather than relying on one-time authorisation decisions that persist indefinitely. The company argues that legacy identity models, built on static policies and standing privileges, were never designed to control agents that invoke tools, access sensitive data and delegate to sub-agents at machine speed.
Built on the SGNL acquisition
Continuous Identity for AI Agents is powered by technology from CrowdStrike’s recent acquisition of SGNL, and assigns every agent a cryptographically verifiable identity based on the open SPIFFE standard, replacing static credentials such as API keys. Access is evaluated contextually each time an agent acts, with that context preserved when an agent delegates work to a sub-agent, and revoked the moment it is no longer needed — a zero standing privilege approach. The capability works alongside Falcon AI Detection and Response, which inspects prompts and intent to flag permission misuse or attempts to push a model beyond its authorised scope.
“Point-in-time authorization becomes a legacy approach the second agents are given autonomy. Authorize once and trust indefinitely is not a security model, it’s a liability,” said Elia Zaitsev, chief technology officer at CrowdStrike.
Agentic MDR gains a marquee customer
The launch coincided with news that Grant Thornton Advisors, the professional services firm, is standardising its managed security services on CrowdStrike’s Falcon Complete platform through the Falcon Complete for Service Providers programme, replacing legacy managed detection and response tools with CrowdStrike’s Agentic MDR. The firm said the move pairs CrowdStrike’s detection and response capabilities with its own advisory and incident response expertise across nearly 20 markets.
Continuous Identity for AI Agents extends CrowdStrike’s risk-aware authorisation across human, non-human and AI agent identities, spanning on-premises, SaaS, browser and cloud environments.
The launch reflects a broader push across the cybersecurity industry to treat AI agent identity as a distinct security category, as enterprises move from experimenting with single AI assistants to deploying fleets of autonomous agents with system-level access to sensitive data and applications.
