AI security guide · MITRE ATLAS
Cross-system agent lateral movement and execution boundaries
How can an AI agent become a bridge between systems for an attacker? Agents connected to email, tickets, code, chat and cloud become bridges: influence gained in one system can be carried into another through the agent’s tools.
Why it matters
Traditional segmentation assumes systems talk through defined integrations. A broadly connected agent is an integration with every system at once, driven by natural language.
How it works
- Influence the agent in a low-trust system (for example a public ticket).
- Agent carries instructions into a high-trust system (cloud, code repository).
- Multi-agent setups can propagate prompts between agents.
Curated agent example
Fictional example: an engineering productivity agent
- The agent reads public issues and can open pull requests and run CI.
- A crafted issue instructs it to add a dependency from an attacker-controlled source.
- The agent opens a PR; CI runs with deployment credentials.
- Mitigation: separate agents per trust zone; no write to code from agents reading public content; human review of agent PRs.
Fictional and simplified for education.
On AWS: architecture notes
- Separate agent runtimes and roles per trust zone (public-facing vs internal).
- Contain agent execution in isolated compute (separate accounts or VPCs) with restricted egress.
- Restrict which agents may call which other agents.
MITRE ATLAS techniques
- AML.T0053 AI Agent Tool Invocation
Execution · Privilege Escalation · Lateral Movement · Maturity: Demonstrated
Adversaries may use their access to an AI agent to invoke tools the agent has access to. LLMs are often connected to other services or resources via tools to increase their capabilities. Tools may include integrations with other applications, access to public or private data sources, and the ability to execute code.
- AML.T0118 Autonomous AI Agent Communication
AI Attack Adaptation · Maturity: Realized
Autonomous AI agents may exchange operational information with other AI agents, sub-agents, or independent agent runs. Exchanged information may include discoveries, objectives, tasking, capabilities, credentials, constraints, operating rules, task state, targeting information, instructions, or results. Communication enables autonomous AI agents to coordinate activities, share discoveries, delegate work, request assistance, validate results, or revise future actions without requiring a human operator to direct each interaction.
- AML.T0143 LLM Prompt Propagation
Lateral Movement · Maturity: Demonstrated
Adversaries may cause malicious instructions already introduced into an AI-enabled environment to be transferred onward to additional storage locations, agents, applications, or resources where they can influence subsequent AI activity. This extends the attack beyond its initial point of introduction and may allow adversaries to influence additional targets without directly delivering a malicious prompt payload to each one.
- AML.T0105 Escape to Host
Privilege Escalation · Maturity: Realized
Adversaries may break out of a container or virtualized environment to gain access to the underlying host. This can allow an adversary access to other containerized or virtualized resources from the host level or to the host itself. In principle, containerized / virtualized resources should provide a clear separation of application functionality and be isolated from the host environment.
ATLAS mitigations
- AML.M0032 Segmentation of AI Agent Components
Define enforceable security boundaries around AI agent tools, data sources, identities, and execution environments. Mediate access through authenticated APIs, isolate code execution via containers or virtual machines, restrict filesystem and network access, and limit tool invocation rates. Build execution environments from clean base images for each run, and do not carry forward any operational state.
- AML.M0037 AI Agent Authority Expansion Controls
Limit an AI agent's ability to autonomously acquire additional authorities that expand its effective permissions during execution. The maximum authority available to the agent should be explicitly granted prior to runtime. Additional resources, identities, services, and targets discovered during execution should be treated as outside the authorized boundary unless they are independently validated and added to scope.
- AML.M0038 AI Agent Scope Drift Detection
Continuously evaluate whether an AI Agent's planned actions remain consistent with its current authorized objective throughout execution. As autonomous agents interact within a dynamic environment, they may discover or generate intermediate objectives or adapt their strategy based on environment feedback.
Related Enterprise ATT&CK techniques
- T1021.007 Remote Services: Cloud ServicesRelated enterprise behavior in the same attack chain (Securitain analysis). Not an ID equivalence.
- T1611 Escape to HostATLAS-published reference: ATLAS publishes T1611 as the ATT&CK reference for this technique.
Controls to prioritize
Restricted agent egress and connectors
Limit the domains and services an agent runtime can reach, avoid generic "fetch any URL" tools, and approve third-party connectors before use.
MITRE ATLAS mitigation: Segmentation of AI Agent ComponentsDedicated least-privilege identity per agent and tool
Give each agent tool its own narrowly scoped role (or a session policy) instead of a shared broad role, so a manipulated agent can only do what that tool legitimately needs.
Security best practices in IAMTreat retrieved and tool-returned content as untrusted
Label provenance, keep untrusted content separate from instructions, restrict which tools can run after untrusted input is read, and control who can write to knowledge sources.
MITRE ATLAS mitigation: Restrict AI Agent Tool Invocation on Untrusted Data
Common misconceptions
- “One powerful agent is simpler and safer to manage.”
- Concentrating connections concentrates risk; several narrow agents are easier to bound.
Checklist
- Map which systems each agent connects
- Split agents by trust zone
- Isolate agent execution environments
Related attack scenarios
Malicious content influencing an agent’s privileged tool invocation
Compromised deployment pipeline role
More AI security guides
References
MITRE ATT&CK® and ATT&CK® are registered trademarks of The MITRE Corporation. ATT&CK data © 2026 The MITRE Corporation. This work is reproduced and distributed with the permission of The MITRE Corporation (ATT&CK v20.1). MITRE ATLAS™ data (v2026.10) © MITRE, used under the Apache License 2.0. Securitain is not affiliated with or endorsed by MITRE. Explanations, scenarios and control mappings on this page are Securitain’s own analysis. Sources and licensing