AI Security - Okta Unveils New Platform for AI Agents Management
Basically, Okta created a tool to help companies manage and secure their AI agents.
Okta has launched a new platform to manage AI agents effectively. This tool aims to enhance security and control access, addressing significant risks. Organizations can now better oversee their AI deployments, ensuring safer operations.
What Happened
On March 16, 2026, Okta unveiled its new platform, Okta for AI Agents, aimed at managing the increasing number of AI agents in organizations. As AI agents become more prevalent, many companies are deploying them without sufficient oversight, leading to potential security risks. A notable incident involved an AI agent mistakenly deleting a live production environment, highlighting the urgent need for better management tools.
Despite the growing adoption of AI agents, a survey revealed that only 22% of enterprises have integrated these agents into their identity management systems. Okta's Chief Product Officer, Ely Kahn, noted a significant increase in the capabilities and numbers of AI agents, with some organizations reporting up to 2,500 agents in production. This exponential growth raises concerns about how security teams can effectively manage and monitor these agents.
Who's Being Targeted
The new platform is designed for organizations of all sizes that are utilizing AI agents. As companies race to implement these agents, they often overlook the risks involved. Okta aims to help these organizations by providing a structured approach to managing AI agents, ensuring they are registered, monitored, and controlled effectively.
With AI agents potentially handling sensitive tasks, the risk of misuse or errors increases. Okta's solution addresses these challenges by integrating AI agents into existing identity management frameworks, allowing organizations to maintain visibility and control over their AI deployments.
Key Features
Okta for AI Agents introduces several key features to enhance security and management:
- Agent Registration: The platform can detect and register AI agents, including those that may be used without official approval (shadow AI).
- Agent Gateway: This central hub controls and monitors agent access to resources, simplifying the management of permissions.
- Ephemeral Access Tokens: AI agents receive temporary access tokens, reducing the risk of long-term token theft, which was a significant issue in previous security breaches.
These features collectively aim to create a secure environment for AI agents, aligning their management with traditional identity management practices.
What to Watch
As Okta continues to develop its platform, it plans to expand beyond identity management into broader AI security areas. The introduction of features like a kill switch for AI agents, which can revoke access if an agent deviates from its intended function, underscores the importance of oversight in AI operations.
Okta's CEO, Todd McKinnon, emphasized the significance of this platform, stating that it might be the most important product they've ever created. As AI technology evolves, so too will the strategies needed to secure it, making Okta's advancements crucial for organizations navigating this new landscape.