iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Virtue AI raised $30 million in seed and Series A funding in April 2025 to develop and expand security products for AI models and autonomous agents. Its tools were designed to test systems for attacks and unsafe behavior, enforce policies as AI operates, and help organizations monitor agent activity. Fortinet announced on August 17, 2026, that it had acquired the company and planned to integrate its capabilities into Fortinet’s AI-Native Security Fabric.
What Virtue AI raised the $30 million to do
San Francisco-based Virtue AI announced the seed and Series A financing in April 2025. SecurityWeek reported on April 15 that Lightspeed Venture Partners and Walden Catalyst Ventures led the round. The company said it would use the money for research and development, market expansion, and hiring in AI safety and security.
The investment addressed a growing operational problem: organizations deploying AI need to assess risks before release and manage them while systems are in use. Virtue AI’s approach combined automated testing, output guardrails, and controls for AI agents rather than focusing only on model evaluation.
Which AI deployment risks Virtue AI targeted
Virtue AI’s products were intended to address risks that can arise in AI models, applications, and agents, including:
#1 Best Overall
- Data poisoning: compromised or manipulated data can undermine a system’s behavior.
- Hallucinations: a model may produce false or unsupported information.
- Prompt injections and jailbreaks: crafted inputs may try to override instructions or safety restrictions.
- Privacy leaks: an AI system may expose sensitive information in its responses or actions.
- Compliance failures: outputs or behavior may conflict with an organization’s rules or regulatory obligations.
These are categories the company said its products addressed, not a guarantee that every attack or failure can be prevented. The reports and company announcements cited here do not provide independent performance testing.
How its products were designed to protect AI systems
Original product lineup
SecurityWeek described Virtue AI’s initial product set as a combination of red-teaming, guardrails, and agent protection. VirtueRed assessed models, systems, and agents; VirtueGuard-Text, VirtueGuard-Image, and VirtueGuard-Video covered guardrails for different content types; and VirtueAgent focused on autonomous agents. SecurityWeek reported that VirtueRed used more than 100 red-teaming algorithms across more than 320 safety categories aligned with regulatory requirements. Those figures describe the product as reported, not an independent benchmark.
Axios characterized the company’s broader offering as continuous algorithmic red-teaming, guardrails intended to block outputs that violate policies or regulations, and protections for autonomous systems. Axios reported customers in finance, healthcare, IT, and frontier labs, naming Uber and Glean. It also reported that Virtue AI planned to grow its team from 20 people to about 50 during 2025.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesAgentSuite and agent-level controls
Virtue AI later described AgentSuite as a unified security stack for autonomous agents. Its announced components covered distinct parts of the agent lifecycle:
Rank #3
- VirtueRed for Agents for continuous red-team testing.
- MCPGuard for vulnerabilities in MCP servers and tools.
- ActionGuard for real-time policy enforcement on agent actions.
- Unified Agent Gateway for routing agent activity through a common control point.
- Observability for conversations, actions, and tool calls, alongside role-based access control and centralized audit logging.
Together, these capabilities were aimed at a problem beyond unsafe text: agents can use tools and take actions, so organizations need to see what they do and apply permissions and policies to those actions. The product descriptions establish Virtue AI’s intended controls, but do not establish their independent effectiveness in production.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What happened to Virtue AI after its funding round
On June 29, 2026, Virtue AI announced that Sohaib Shaikh had become CEO and Daniel Le CFO/COO, following the departures of co-founders Bo Li and Sanmi Koyejo. In that company release, Virtue AI reported 600 percent year-over-year growth and said it had launched AgentSuite, Agent ForgingGround, PolicyGuard, and Shadow AI in 2026. The growth figure and product launches were company-reported.
On August 17, 2026, Fortinet announced that it had acquired Virtue AI, with plans to integrate the company’s capabilities into Fortinet’s AI-Native Security Fabric. Fortinet did not disclose financial terms and described them as immaterial to Fortinet. The announcement positioned the acquisition as extending protection across AI models, applications, and agentic systems from development through runtime.
Fortinet described the acquired capabilities as including agent red-teaming in more than 50 sandboxed environments and 14 high-stakes domains; visibility and governance for agents and tools; continuous validation across model updates and policy fine-tuning; automated red-teaming across hundreds of attack vectors and more than 1,000 risk categories; and real-time guardrails for text, images, video, audio, and AI-generated code. These are figures in Fortinet’s acquisition announcement; they should not be compared directly with VirtueRed’s earlier 100-plus algorithm and 320-plus category figures, which describe a different product scope and reporting context.
Best Value
Fortinet’s announcement also cited Gartner’s forecast that products and tools for securing AI ecosystems and AI agents could grow from $2.8 billion in 2026 to $16.4 billion by 2030. That is a forecast attributed to Gartner by Fortinet, not a measured market total.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
