[[[SUMMARY_START]]]
A cybersecurity startup focused on securing autonomous AI agents has raised $190 million, according to a report dated March 10, 2026.
The company was founded by the founder of Mandiant, a firm known for incident response and threat intelligence work.
The funding highlights growing investor attention on security tools designed for AI systems that can take actions on behalf of users.
Details on the investors, valuation, and product rollout were not disclosed in the available information.
[[[SUMMARY_END]]]
A cybersecurity startup founded by the founder of Mandiant has raised $190 million to build security technology aimed at autonomous AI agents, according to a report published Tuesday.
The financing round underscores increasing interest in tools designed to protect a new class of software: AI agents that can plan tasks, call external tools, and execute actions with limited human oversight. As organizations experiment with agentic AI for customer support, software development, IT operations, and business workflows, security teams are assessing how to manage risks that differ from those posed by traditional applications.
The report did not provide the startup’s name, the identities of participating investors, the company’s valuation, or the structure of the financing. It also did not specify whether the $190 million was raised in a single round or across multiple tranches.
Mandiant’s founder is widely associated with building incident response capabilities and threat intelligence operations that helped shape modern enterprise security practices. The new startup’s focus on autonomous AI agents reflects a shift in the security market toward protecting systems that can make decisions and take actions, rather than only storing or processing data.
## Security concerns around autonomous AI agents
Autonomous AI agents differ from conventional software in that they may be granted permissions to access internal systems, retrieve data, and perform tasks such as sending messages, modifying records, or initiating transactions. Security teams have raised concerns that these capabilities can expand the potential impact of errors, misuse, or compromise.
In enterprise settings, agentic systems may interact with email, calendars, customer relationship management platforms, code repositories, ticketing systems, and cloud infrastructure. If an agent is misconfigured or manipulated, it could take unintended actions at scale. Security controls for such systems typically need to address identity and access management, authorization boundaries, logging, and monitoring of agent actions.
Another challenge is that AI agents often rely on external tools and integrations. Each integration can introduce additional attack surfaces, including API keys, tokens, and third-party services. Security products aimed at this space may need to track how agents obtain credentials, what permissions they hold, and how they use them over time.
The report did not detail the specific technical approach the startup is taking. However, the stated focus on autonomous AI agent security suggests the company is positioning itself in a segment where organizations are seeking ways to govern agent behavior, enforce policy, and detect anomalous activity.
## Funding signals investor focus on agentic AI risk
The $190 million raise, as reported, is a large sum for a security startup and indicates strong investor appetite for companies addressing emerging AI-related risks. In recent years, enterprises have increased spending on cloud security, identity security, and data protection, while also evaluating new controls for AI deployments.
Autonomous agents can blur traditional security boundaries because they may operate across multiple systems and act on behalf of users or teams. That can complicate accountability and auditing, particularly when actions are generated through model-driven reasoning rather than deterministic code paths.
Security leaders have also emphasized the need for clear governance over AI systems, including policies for what agents are allowed to do, how they are tested before deployment, and how incidents are investigated when something goes wrong. Tools that provide visibility into agent decisions and actions, along with mechanisms to constrain behavior, are increasingly being discussed as requirements for broader adoption.
The report did not indicate how the startup plans to allocate the new capital, such as toward product development, hiring, go-to-market expansion, or acquisitions.
## Founder background and market positioning
Mandiant’s founder is known for building a company that became prominent for responding to major cyber incidents and tracking advanced threat activity. That background may be relevant to the new venture’s positioning, as organizations deploying AI agents are likely to demand incident response readiness, forensic visibility, and threat-informed defenses tailored to AI-driven workflows.
The report did not provide a timeline for product availability, customer deployments, or partnerships with AI platform providers. It also did not specify whether the startup is targeting large enterprises, mid-sized organizations, or regulated sectors such as finance and healthcare.
As agentic AI capabilities expand, security vendors and startups are competing to define standards for safe deployment, including how to authenticate agents, limit privileges, validate tool use, and monitor actions. The reported $190 million raise suggests the new company intends to play a significant role in that market, though further details about its technology and commercial plans were not included in the available information.
