Teleport has announced a significant expansion of its Identity Security platform, introducing three new features designed to manage and secure the behavior of autonomous AI agents within production infrastructure. These additions, Beams Session Summaries, Agentic Classifiers, and Risk Scoring, aim to establish a framework for identifying and preventing agent misalignment.
The new capabilities are integrated with Beams, Teleport’s trusted runtime environment for agents. The company recently published a white paper, "From Zero Trust to Agent Trust," which argues that traditional zero trust principles, while necessary, are insufficient for governing agents operating at scale. This paper extends zero trust into three agent trust principles: "Verify explicitly" becomes "Enforce continuously," emphasizing a unique, attestable identity and operation within a trusted runtime; "Use least privileged access" evolves into "Bound collective autonomy," addressing the potential for individually authorized actions by a swarm of agents to become collectively destructive; and "Assume breach" transforms into "Assume misalignment," advocating for continuous monitoring and real-time intervention to detect and respond to agent drift.
Beams Session Summaries provide a concise, human-readable overview of an AI agent's actions, including its identity, privileges, tool and API calls, and LLM prompts and responses. This summary also includes the agent's reasoning, establishing a behavioral baseline against its declared objective.
Agentic Classifiers enable the creation of policies that evaluate the behavior of individual agents or groups of agents against company-specific criteria. This allows for the flagging of agent actions that deviate from their stated objectives.
Risk Scoring automatically summarizes SSH, Kubernetes, and database sessions, categorizing them by risk level and mapping actions to the MITRE ATT&CK framework. This feature allows infrastructure and security teams to automate or manually search across sessions for specific commands, resources, or behaviors.
According to Teleport, these new capabilities, in conjunction with Beams, lay the foundation for agent trust by providing agents with a cryptographic, continuously monitored identity. They also make individual and collective risk visible before actions are taken and equip enterprises with tools to detect and respond to misalignment in real time, operationalizing the "assume misalignment" principle. The company states that these features provide an operational harness to observe agent actions, classify expected behavior, and assess the risk of future actions.
