Implementation Guide

How to Govern AI Agents Handling Export-Controlled Technical Data

Governing AI agents that handle ITAR- or EAR-controlled technical data requires runtime enforcement, not static policy. This means enforcing attribute-based access checks on every tool call, giving each agent a distinct auditable identity, propagating export control classification through retrieval pipelines, and logging the full chain of agent actions so compliance can be demonstrated after the fact rather than assumed in advance.

Architectural Components for Runtime Enforcement

Effective governance for agents handling controlled technical data depends on a small set of runtime components working together, rather than on policy documentation alone.

  1. 01

    Attribute-Based Access Checks on Every Tool Call

    Each time an agent invokes a tool or retrieves data, the request is evaluated against attributes such as user nationality, clearance level, and project assignment before it is permitted.

  2. 02

    Distinct, Auditable Agent Identity

    Agents operate under their own identity rather than inheriting a shared service account, so every action can be traced to the specific agent and invoking user.

  3. 03

    Classification Propagation Through Retrieval Pipelines

    Export control classification attached to source documents must travel with that data through embeddings, retrieval, and generated outputs, not just sit in a static access list.

  4. 04

    Full Chain-of-Action Audit Logging

    Every retrieval, tool call, and handoff an agent performs is logged in enough detail to reconstruct the full action chain during a compliance review.

Where Static Controls Fail Against Agent Behavior

Traditional data loss prevention and access control models assume predictable, human-initiated query patterns. Agent behavior breaks those assumptions in several specific ways.

Dynamic Tool Calls

Agents invoke tools and data sources unpredictably, outside the fixed query patterns that DLP systems expect.

RAG Retrieval Gaps

Vector search can surface controlled technical data into a context window without document-level permission checks.

Agent-to-Agent Handoffs

Controlled data can propagate across sessions or systems with no persistent chain-of-custody record.

No Agent-Specific Rules

ITAR and EAR do not yet define agent behavior explicitly, forcing enterprises to extrapolate from deemed export and encryption exemption rules.

Governance Decisions to Make Before Deployment

Before deploying an agent against data that may fall under ITAR or EAR, the following decisions should be resolved and documented.

  • Confirm which datasets meet ITAR technical data or EAR technology definitions, including derivative outputs an agent might generate.
  • Determine whether agent infrastructure, including model hosting, could be accessible to foreign persons, triggering deemed export analysis.
  • Define attribute-based access conditions (nationality, clearance, project assignment) that must be checked before an agent can retrieve or transmit controlled data.
  • Establish key management practices that separate data retrieval from decryption authority for agent-initiated requests.
  • Identify where human-in-the-loop review is required for borderline classification decisions automated checks cannot resolve confidently.
  • Specify what an audit log must contain to reconstruct an agent's full action chain for a compliance review.

Why Export Control Compliance for AI Agents Is a Runtime Problem

Export control regimes such as ITAR and EAR were built around predictable, human-mediated transfers of technical data: a document sent, a briefing given, a system accessed by a known individual. AI agents do not fit that model. An agent can retrieve, summarize, translate, and forward controlled technical data across dozens of tool calls in seconds, often in ways that were never explicitly authorized by a human reviewer at the time. Static policy documents describe what should happen; they cannot inspect a tool call before it executes. That gap is why compliance for agentic systems has to be enforced at runtime, at the moment each action occurs, rather than assumed from a written policy.

What Counts as Technical Data When an Agent Is Involved

ITAR and EAR definitions of technical data and technology extend to derivative materials, not only original source documents. When an agent summarizes a controlled engineering specification, translates it, or incorporates it into a generated response, that output can itself carry export control obligations. Enterprises need a clear, documented view of which datasets meet these definitions, and they need that classification to be evaluated continuously, since an agent may combine unrestricted inputs in a way that produces a controlled output.

Derivative Outputs Are Often Overlooked

Classification efforts frequently focus on source documents and miss the outputs an agent generates from them. A summary, a translated excerpt, or a synthesized answer drawing on controlled material can carry the same restrictions as the original, even though it was never separately reviewed.

The Encryption Exemption Model and What It Requires of Agent Systems

Some export control frameworks permit the transfer of technical data across borders when it remains properly encrypted and access to decryption is controlled. Applying that model to agent systems requires separating the ability to retrieve data from the ability to decrypt it. An agent, or the infrastructure running it, should not hold blanket decryption authority; decryption should depend on the same attribute checks (nationality, clearance, project assignment) applied to any other access request. This distinction between retrieval and decryption is a key architectural requirement, not a compliance formality.

Agent Identity and Least Privilege in Practice

Agents that operate under a shared service account or inherit a user's full permission set make it difficult to answer basic compliance questions: which agent accessed this data, on whose behalf, and under what authorization. Giving each agent a distinct, auditable identity, scoped to the minimum permissions its task requires, makes it possible to enforce least privilege at the point of action and to reconstruct exactly what happened after the fact. This is the same principle underlying least-privilege access control for human users, applied to non-human actors that can act far faster and across more systems than a person typically would.

Tradeoffs and Open Questions

Runtime enforcement introduces real tradeoffs. Attribute checks on every tool call add latency and complexity to agent workflows. Human-in-the-loop review, while necessary for borderline classification decisions, slows down the automation gains that motivated agent deployment in the first place. And because ITAR and EAR do not yet define agent behavior explicitly, enterprises are extrapolating from deemed export and encryption exemption rules written for a different technical context. Reasonable organizations may reach different conclusions about where to draw these lines, which is why documented decisions and defensible audit trails matter as much as the technical controls themselves.

Runtime Enforcement Is the Control That Documentation Cannot Provide

Trussed AI provides runtime governance and security for enterprise AI agents, including agent identity, permission scoping, tool approval workflows, and audit logging built for the way agents actually operate.

Request a Demo