AI Agent Security Takes Center Stage as Obsidian Raises $85M to Protect SaaS

SaaS

AI Agent Security is becoming critical as Obsidian raises $85M to control AI agents modifying, deleting and accessing sensitive data across SaaS applications.

Obsidian Security raised $85 million at a $1.1 billion valuation to expand AI security.
Its platform protects SaaS applications and sensitive enterprise data from risky AI agent actions.
More than 65% of protected enterprises have given AI agents access to third-party application data.
Obsidian is developing real-time monitoring and controls to stop dangerous agent activity.
The funding highlights a growing challenge: AI agents are gaining access faster than many businesses can secure them.

AI Agent Security is becoming a pressing enterprise issue as Obsidian Security raises $85 million in Series D funding at a $1.1 billion valuation. The company wants to help businesses control AI agents that now have direct access to sensitive data inside SaaS applications.

Crescent Cove Advisors led the funding round. The investment will help Obsidian expand security controls across platforms including Google Drive, Notion, Slack, Salesforce, Snowflake, Databricks, GitHub and GitLab.

What happens when an AI agent has permission to change or delete critical business data?

That risk is already emerging. Obsidian CEO Hasan Imam says more than 65% of the enterprises it protects have given AI agents access to data inside third-party applications. These systems can do more than retrieve information. They can modify records, delete content and carry out tasks on behalf of users.

Traditional security often focuses on who has access. AI agent security for SaaS must also track what an agent does once it enters an application. Obsidian already monitors AI agents connected to Claude Code, ChatGPT, Copilot and Salesforce Agentforce. It plans to extend protection to more platforms as companies expand their use of autonomous systems.

The concern becomes stronger when basic protections are still missing. Obsidian says 75% of applications have an administrator without multifactor authentication. Some businesses also continue to use local accounts instead of centralized identity controls.

For companies deploying AI agents in SaaS applications, a mistake can have serious consequences. An agent with excessive permissions could change sensitive records, remove repositories or cause major disruption.

Obsidian believes security teams need to see and control agent actions in real time. Instead of waiting for an alert after an incident, security systems should be able to stop dangerous activity as it happens. AI agents can deliver greater automation and productivity, but their growing access also creates a larger security surface.

Obsidian’s funding highlights a clear market shift. Businesses stopped asking only what AI agents can accomplish. They are increasingly asking what those agents should be allowed to do. Business Honor asks Are enterprises ready to give AI agents greater control before they have greater control over the AI itself?