A cluster of identity and security vendors has rolled out tooling for governing AI agents, converging on a shared idea: a static, one-time access decision is not enough once an agent can act on its own, so authorization has to be continuous, aware of context and intent, and tied back to an accountable human or owner. The launches, many timed to the Identiverse conference, mark agentic identity moving from concept to product.
CrowdStrike introduced Continuous Identity for AI Agents as part of its Falcon Next-Gen Identity Security line, authorizing each agent action in real time against the agent’s owner, the calling identity, device risk posture, and live risk signals. The capability gives agents a cryptographically verifiable identity using the SPIFFE standard and preserves ownership data through chains of sub-agent delegation, on a zero-standing-privilege model. “AI agents are transforming how work gets done, and how identities must be secured. Point-in-time authorization becomes a legacy approach the second agents are given autonomy,” said Elia Zaitsev, Chief Technology Officer of CrowdStrike. The technology comes from CrowdStrike’s acquisition of SGNL.
SailPoint is buying its way deeper into the same problem, announcing an intent to acquire the Tel Aviv-based non-human identity firm Entro. SailPoint plans to fold Entro into its Agentic Fabric platform, adding discovery and credential security spanning more than 1,000 non-human identity and agent types and 1,200 credential types across 70 enterprise sources, along with detection and response for machine identities. The deal is expected to close in the third quarter of SailPoint’s 2027 fiscal year.
Beyond Identity launched Ceros, which it calls a trust layer purpose-built for autonomous AI agents, providing identity, observability, and governance in one platform. Ceros applies a five-dimensional identity check spanning the agent’s operator, how the user behind it was authenticated, device compliance, the applications and tools the agent uses, and the agent’s own configuration, alongside discovery, runtime policy enforcement, and audit trails. “CISOs get complete visibility and control across every agent, identity, device and action with a platform built for autonomous AI at scale,” said Jasson Casey, Chief Executive Officer of Beyond Identity.
Akamai approached agentic identity from the network edge, unveiling an agentic security framework for its Bot and Agent Control work that ties together agent identity, permissioning, traffic analysis, and edge enforcement. The framework draws in Visa’s Trusted Agent Protocol for permissioned transactions, a Know Your Agent effort with Skyfire and Experian to verify an agent’s identity and origin, and integrations with Auth0 and Ping Identity to apply user-centric authentication to agents. Akamai framed the task as a dual-identity problem: recognizing not just the agent, but the person behind it.
Saviynt rounded out the wave by expanding its Agent Access Gateway with Intent-Aware Runtime Authorization, evaluating an agent’s actions in real time against identity, context, policy, and intent and blocking those that fall outside it. The release adds identity verification features, including biometric scanning and liveness detection spanning more than 177 countries, meant to reduce the risk of agent impersonation. The moves echo broader industry work such as Ping Identity’s Identity for AI to govern the autonomous agent economy.
Across the five launches, the common thread is a shift from authenticating an agent once to continuously authorizing what it does, and insisting that every agent trace back to a human or owner who can be held accountable.
Sources: CrowdStrike, SailPoint, Beyond Identity, Akamai, Saviynt
–
By Ali Nassar-Smith
