Thousands of AI agents are operating inside corporate systems without anyone tracking them, creating a growing security problem that identity and data-security vendors such as Cyera, Okta and SailPoint are racing to address.
AI agent governance · Enterprise security
Thousands of AI agents are running inside companies, and most are not tracked
AI agents can read files, call applications and make changes on a user's behalf. Without a central inventory of who approved them and what they can access, security teams are facing a growing blind spot that vendors such as Cyera, Okta and SailPoint are racing to close.
Thousands
AI agents reported operating inside corporate systems, a scale at which manual oversight becomes difficult.
3
Vendors with different starting points now moving toward the same agent governance problem.
CONVERGENCE
Three starting points, one problem
Cyera
Data security posture: locating sensitive data and who or what can reach it.
Okta
Identity, expanding to non-human identities such as service accounts and automated credentials.
SailPoint
Identity, expanding to non-human identities used by agents.
Agent governance is emerging as a distinct product category, not just a feature of existing tools.
BASELINE CONTROLS
What observers agree agents need
1. Inventory
Know what each agent is and who approved it.
→
2. Least privilege
Limit each agent to the systems and data it needs.
→
3. Auditable logs
Record every action an agent takes.
OPEN QUESTIONS
The gap is clear. The scale beyond the headline is not.
Affected companies
No detailed figures on how many organizations are affected.
Incidents
No specific incidents have been detailed.
Speed of adoption
Whether vendors and enterprises can keep pace with fast-moving deployments.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article...