TL;DR: Cyera’s acquisition of Ryft extends its agentic AI security platform toward governed data access, as the company says autonomous systems need trusted, traceable data controls and visibility into identities and agents accessing data. The shift shows that data governance and identity context are converging into one control problem when AI systems act on enterprise data.
At a glance
What this is: Cyera’s purchase of Ryft is framed as a move to tighten agentic AI data governance by linking governed data access with identity and agent visibility.
Why it matters: IAM, IGA, and security teams should read this as a sign that agent governance is moving from model-level controls to data-access controls that must track identity, delegation, and task context.
By the numbers:
- Cyera says it recently raised a $400 million Series F at a $9 billion valuation.
Context
Agentic AI data governance is the discipline of controlling what data AI systems and agents can reach, how that access is traced, and what happens after the access occurs. In this deal, Cyera frames the problem as one of governed data access for systems that do not behave like traditional users.
The important governance gap is that autonomous or semi-autonomous agents can change identity, tools, and execution paths during a task, which means static access views no longer describe the real risk. For identity programmes, this pushes data security, delegation tracking, and access context into the same operating model.
The acquisition also signals that the market is converging on control planes that connect data, identity, and agent behaviour rather than treating them as separate review queues. That matters because agentic AI introduces a moving target that conventional review cadences struggle to capture.
Key questions
Q: How should teams govern AI agent access to production data?
A: Treat agent actions as runtime events that need policy checks at execution time, not just pre-approved credentials. High-risk reads, writes, and deletes should be bounded by context, session limits, and approval gates that apply before the command completes.
Q: Why do agent delegation chains create data governance risk?
A: Delegation chains create risk because the effective actor may change as work moves across agents, tools, and approvals. That breaks the assumption that one account maps cleanly to one purpose. If the chain is not visible in logs and policy, teams cannot tell which actor actually consumed or acted on governed data.
Q: What breaks when agentic AI is governed like a normal application account?
A: Security controls break down because agentic systems do not behave like fixed-function applications. They can choose actions at runtime, combine tools in unexpected ways, and move faster than periodic review cycles. That means static roles, annual recertification, and one-time approvals do not fully describe the risk or contain the behaviour.
Q: How should security teams classify AI agents before writing controls?
A: Start by asking whether the system only responds, suggests with human approval, or executes on its own credentials. That classification determines whether chatbot safeguards, copilot review, or NHI governance is the right control model. If you skip classification, you will either over-control harmless assistants or under-control autonomous systems.
How it works in practice
Why agentic AI data governance needs identity context
Agentic AI data governance is not only about data classification. It also depends on knowing which identity, delegation chain, or task context is acting on the data at runtime. Cyera’s framing points to the difference between a user opening a dataset and an agent traversing multiple tools, identities, or execution steps to reach the same data. That changes auditability, access scoping, and post-access accountability. In practice, the governance question becomes whether data controls can describe the actor’s effective authority at the moment of use, not just at provisioning time.
Practical implication: Map agent data access to the identity and delegation context that produced it, not just to the data asset itself.
Secure data lakes for AI agents
A secure and automated data lake for AI agents implies a layer that mediates access, traces consumption, and preserves control over downstream use. The architectural issue is not storage alone, but whether the data plane can enforce policy when agents query, combine, and act on information across workflows. For agentic systems, the control point has to move closer to where data is consumed because agent behaviour can be task-specific and dynamic. That makes lineage, entitlement context, and access evidence part of the security boundary.
Practical implication: Treat AI-agent data stores as governed execution environments, not passive repositories.
Why delegation chains complicate traceability
Cyera’s own description highlights a key reality of agentic systems: agents can shift identity based on task, tools, and chain of delegation from other agents. That means the effective actor may not be a single stable identity, but a sequence of delegated actions with different scopes and purposes. Traditional identity controls often assume a stable subject and a stable review unit. Once delegation becomes chained and dynamic, traceability must follow the actor path across handoffs, not just the nominal account at the start or end of the workflow.
Practical implication: Build logging and review around delegation paths so investigators can reconstruct who or what actually touched the data.
NHI Mgmt Group analysis
Agentic AI data governance is becoming a control-plane problem, not a storage problem. The Cyera-Ryft deal reflects a market shift away from treating AI data protection as a classification or lakehouse issue. Once agents can query, combine, and act on data across workflows, the governing question is how identity, delegation, and data access are bound together in the same decision path. Practitioners should expect their data security programme to absorb identity context as a first-class input.
Agent identity is no longer sufficient without task context. Cyera’s description of agents shifting identity based on task, tools, and delegation chain points to a deeper governance change. A static entitlement record does not explain what an agent is allowed to do mid-execution, which means access governance must account for runtime context as well as provisioning state. The implication is that traditional approval and review models need to be re-read through agent behaviour, not just account ownership.
Traceability is now an assurance requirement for agentic AI. If organisations cannot reconstruct which identity, tool path, and data set intersected during an agent task, they cannot prove governed access after the fact. That is especially important where AI agents operate in hybrid estates with mixed human, service, and autonomous controls. The practitioner conclusion is simple: agentic AI governance has to produce evidence, not just policy.
Delegation-chain visibility is the named concept this market is converging on. The article’s central risk is that the real actor is often a chain of delegated identities, not a single user or service account. That breaks governance assumptions built around stable subjects and one-step authorization. The implication is that identity programmes must be able to describe authority across the full chain of delegation, or their controls will only describe the wrong actor.
Data control without agent control will remain incomplete. The acquisition suggests that the next generation of AI security platforms will be judged on whether they can connect data access, identity visibility, and agent behaviour in one operating model. For practitioners, that means separating data governance from identity governance is becoming less defensible. The practical conclusion is that control design must follow the agent, not just the dataset.
From our research library:
- 54% of organisations are actively deploying AI agents across workflows, yet only 21% report a mature governance model for agentic AI.
- Read next: Agentic AI Identity Guide
What this signals
Delegation-chain visibility: agentic AI programmes now need evidence that shows how authority moved across tasks, tools, and delegated identities. When the actor can shift scope during execution, the control surface moves from provisioning to runtime traceability.
Access review cadences will not be enough on their own. Access reviews assume a privilege lasts long enough to be observed and certified. Agentic systems can acquire, combine, and release access inside one task, so governance has to capture the event as it happens.
According to the 2026 Infrastructure Identity Survey, 54% of organisations are actively deploying AI agents across workflows, yet only 21% report a mature governance model for agentic AI. That gap means programme design is still ahead of control design.
For practitioners
- Define the agent data-access boundary Map which datasets AI agents may reach, which identities they inherit, and which delegation paths are allowed to extend that access.
- Instrument runtime identity and data lineage Capture which agent, which tool, and which dataset intersected during each task so investigators can reconstruct effective authority.
- Review delegation chains for scope drift Test whether an agent can change identity, tool use, or data reach during a task without a new authorization checkpoint.
- Align access reviews to evidence, not ownership Require review artefacts that show actual agent data use, because account ownership alone does not describe what the system did.
- Separate governed analytics from unrestricted exploration Create distinct policies for production data, model experimentation, and agentic workflows so broad data access does not become the default.
Key takeaways
- Agentic AI data governance is shifting from static data protection to runtime control over identity, delegation, and data use.
- The Cyera-Ryft deal signals that security teams increasingly need traceability for how agents move across data and tools, not just where data is stored.
- Practitioners should design governance that can explain effective authority after a task completes, or agentic behaviour will outrun review cycles.
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
OWASP Agentic AI Top 10, OWASP Non-Human Identity Top 10 and MITRE ATT&CK address the attack and risk surface, while NIST AI RMF sets the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| OWASP Agentic AI Top 10 | ASI03 — Identity & Privilege Abuse | The article centers on agents shifting identity and scope during delegated data access. |
| ASI07 — Insecure Inter-Agent Communication | Delegation chains and agent handoffs make inter-agent trust and traceability central here. | |
| Recommendation — Apply ASI03 controls to constrain privilege changes across delegated AI workflows. Harden inter-agent handoffs and verify authority before passing data between agents. | ||
| OWASP Non-Human Identity Top 10 | NHI-05 — Overprivileged NHI | Agent data access depends on whether non-human identities can exceed their intended scope. |
| Recommendation — Review agent-associated non-human identities for excess data access and trim unused scope. | ||
| NIST AI RMF | GOVERN — AI Governance and Accountability | This deal is about building governance around agentic AI and the accountability model it needs. |
| Recommendation — Define governance and accountability for AI data access decisions before agents scale further. | ||
| MITRE ATT&CK | TA0006;TA0008 — Credential Access; Lateral Movement | The core threat pattern is delegated access that expands into broader data movement and use. |
| Recommendation — Map agent data traversal paths to Credential Access and Lateral Movement to spot scope creep. | ||
Key terms
- Agentic Data Governance: Agentic data governance is a model where intelligent systems help validate, enrich, route, and repair data in motion instead of waiting for humans to intervene. It aims to keep controls active at pipeline speed, but it still requires clear authority limits, logging, and ownership.
- Delegation Chain: A delegation chain is the sequence of identities, credentials, and tool calls an agent uses to complete a task across systems. It matters because each step may appear acceptable on its own while the combined path produces an outcome no reviewer would have approved directly.
- Identity Traceability: Identity traceability is the ability to link each action back to a specific identity, authorisation path, and time window. It is essential when humans, service accounts, and AI agents all operate in the same environment and auditors need a defensible record.
- Task-Scoped Access: Task-scoped access is permission granted for one defined purpose and removed once the task is complete or the session expires. For non-human identities, it reduces standing privilege and limits how long an attacker can exploit a stolen credential.
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.
Published by the NHIMG editorial team on June 2, 2026.
Updated on October 8, 2026.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org