Vijil Launched Automated Red Teaming for AI Agents
New software uses adversarial agents to identify vulnerabilities in enterprise AI deployments.
Updated on Oct. 7, 2026 in Artificial Intelligence
Live Poll
Do you trust that large companies are sufficiently securing their AI systems against security threats?
Vijil has released the Diamond Adaptive Red Teaming (DART) system, a tool designed to automate security testing for enterprise AI agents. The software identifies policy violations and security gaps by conducting multi-turn adversarial attacks.
Why it matters
As Fortune 500 companies scale from managing a handful of AI agents to thousands, manual security testing has become a bottleneck for engineering teams. DART automates this process to keep pace with evolving attack tactics against large agent fleets.
DART demonstrated a 1.5x higher attack success rate compared to the leading competitor across 9 of 12 specific tasks. The system identifies security vulnerabilities and policy violations by deploying automated agents that perform multi-turn adversarial probes.
The players
Vijil
A cybersecurity firm that specializes in AI security and the development of adversarial testing tools for enterprise agents.
The details
DART employs multiple adversarial agents—autonomous software programs designed to simulate malicious activity—to probe target systems with multi-turn attacks. The system continuously evaluates how the target agent responds, allowing it to adapt its tactics across subsequent turns and episodes. This automated loop allows the software to test defenses more dynamically than static scanners. The platform integrates directly with existing agent frameworks and can be deployed within a customer's Virtual Private Cloud or on-premises.
Timeline
2025: Fortune 500 companies utilized an average of fewer than 15 AI agents.
October 07, 2026: Vijil released the DART system for enterprise use.
2028: Fortune 500 companies are projected to utilize more than 150,000 agents.
The Tech Race
This release directly addresses the security challenges posed by the rapid proliferation of autonomous agents in corporate environments. It follows the trend of Gartner's agent utilization projections for Fortune 500 companies, which anticipate a shift from 15 agents per company today to 150,000 by 2028.
Enterprise customers using Vijil Diamond can access DART immediately at no additional cost. The tool is designed for integration into standard AI agent deployment platforms and private cloud environments.
The takeaway
As enterprise agent fleets grow, automated, adaptive red teaming is becoming a baseline requirement for maintaining system integrity. Engineering teams should monitor whether this benchmark performance holds as new, more complex agent architectures emerge in the coming year.
Further reading
For more on how organizations are securing large-scale deployments, visit our section on Artificial Intelligence.
Live Poll
Do you trust that large companies are sufficiently securing their AI systems against security threats?
