Security platform AEGIS has launched a comprehensive suite designed to protect AI agents and infrastructure against seven critical attack vectors, including prompt injection, RAG poisoning, and supply chain vulnerabilities.

Key Points

• AEGIS provides security across seven vectors, including Model Context Protocol (MCP) attacks, agent hijacking, and vector database breaches.
• The platform features a three-classifier ensemble that blocks prompt injection attempts in under 7 milliseconds with 0.98 confidence.
• It introduces an Identity and Access Management (IAM) engine for AI agents, enforcing default-deny policies on all tool calls.
• Deployment options include fully managed cloud services, Docker containers, and air-gapped Kubernetes environments for regulated industries.
• The company has been selected as the security partner for the proposed NHS ID Card System, a project targeting significant public sector efficiencies.

Why it Matters

As AI agents gain autonomy, securing the interface between models and external tools becomes a critical requirement for enterprise and government adoption. By consolidating protection for multiple attack surfaces into a single platform, AEGIS aims to reduce the complexity of securing AI-driven workflows.