Trust3 AI launches research hub for agent security

Trust3 AI on August 5 launched Trust3 AI Labs, a technical research hub focused on securing AI agents across discovery, observability, runtime enforcement, protocol security, and governance. The move targets the growing attack surface created as enterprises deploy more agents into production systems.

Why it matters: - AI agents are moving into production systems faster than security controls are adapting. - Trust3 AI Labs is designed to study the threats, controls, and architectures needed to reduce risk across the full agent lifecycle. - The work targets systems where reasoning, access, and action happen in one runtime, raising the stakes for enterprise data and application security.

What happened: - Trust3 AI on August 5 announced the launch of Trust3 AI Labs in San Francisco. - Trust3 AI Labs is a dedicated technical research hub for the agentic enterprise. - The Labs will focus on five pillars: agent discovery, full-fidelity observability, runtime enforcement, MCP and A2A security, and unified governance. - Trust3 AI said the Labs will publish research across four areas: Threat Research, Tools, Engineering, and Talks.

The details: - Agent discovery will target both registered and shadow agents across Amazon Bedrock, Microsoft Copilot Studio, Databricks, LangChain, SaaS platforms, internal APIs, and custom SDKs. - Full-fidelity observability will capture and replay complete agent traces, including Prompt, Retrieval, Reasoning, Tool Selection, Tool Call, Response, and Action. - Runtime enforcement will include controls inside the agent decision loop, kill switches, credential revocation, and purpose drift detection. - MCP and A2A security research will cover MCP server discovery, tool provenance, authentication and authorization, tool description poisoning, cross-server trust, and agent-to-agent identity propagation. - Unified governance will connect the other pillars and support continuous policy enforcement across frameworks, platforms, and clouds. - Trust3 AI Labs will share threat research, technical experiments, security tools, reference architectures, and engineering insights with engineers, architects, and security practitioners. - Trust3 AI said the platform helps organizations discover every agent, observe every decision, secure every action, and govern every connected data source across frameworks, platforms, and clouds. - The company said the platform is built on Privacera's production-grade policy enforcement heritage. - Trust3 AI also said the platform combines agent discovery, full-fidelity observability, runtime security, policy intelligence, native enforcement, and continuous governance. - For more information, visit the company's announcement. - Trust3 AI lists a LinkedIn page at company social media.

Between the lines: - The launch signals that agent security is shifting from a feature inside a product to a research problem across the stack. - The focus on shadow agents, tool poisoning, and identity propagation suggests enterprises are now dealing with hidden and interconnected risks, not just isolated model prompts. - By covering both technical research and practical tools, Trust3 AI is positioning the Labs as part threat-intelligence effort and part engineering resource.

What's next: - Trust3 AI Labs will continue publishing research, tools, and technical guidance for teams building and defending agentic systems. - The company is likely to use the Labs to shape how enterprises think about agent discovery, monitoring, enforcement, and governance as adoption expands. - As multi-agent deployments grow, the security questions highlighted by Trust3 AI Labs are expected to become more central to enterprise AI operations.