Execution

T1698: Indirect Prompt Injection

Adversaries may deliver malicious prompts to an LLM indirectly by embedding the prompts in content that the adversary expects may be processed by an LLM.(Citation: NIST indirect prompt injection) Thes...

T1698 · Technique ·2 platforms

Description

Adversaries may deliver malicious prompts to an LLM indirectly by embedding the prompts in content that the adversary expects may be processed by an LLM.(Citation: NIST indirect prompt injection) These prompts can be included in web pages,(Citation: Palo Alto Indirect Prompt Injection ITW) documents, emails, tickets,(Citation: Snyk Indirect Prompt Injection) code, or other sources that an agent, copilot, or other AI system may retrieve and process. When an LLM processes this content, the embedded instructions may cause the system to perform unintended behaviors.(Citation: GTIG AI Threats Sep 2026)

Adversaries may use indirect prompt injection to Impair Autonomous Defenses, and for unauthorized execution of any capabilities available to the LLM.

Platforms

WindowsSaaS

Mitigations (6)

Restrict File and Directory PermissionsM1022

Restrict AI agents to files, directories, repositories, and content stores required for their approved workflows. Limit write access separately from read access where possible to reduce the actions available following successful indirect prompt injection.

Privileged Account ManagementM1026

Prevent AI agents processing untrusted or externally controlled content from automatically exercising privileged permissions. Require additional authorization for sensitive operations derived from retrieved content.

Execution PreventionM1038

Prevent instructions contained in retrieved or externally supplied content from independently authorizing consequential agent actions. Require tool/API requests derived from external content to satisfy authorization and policy controls independent of the model-generated decision.

Network SegmentationM1030

Restrict network resources and external destinations accessible to AI agents that process untrusted content. Limit outbound communication and access to unrelated internal services to reduce exfiltration and lateral effects following successful injection.

User Account ManagementM1018

Use dedicated identities for AI agents and restrict their permissions to the minimum resources and actions required by the approved workflow. Do not allow retrieved content to change the authorization context associated with the agent.

User TrainingM1017

Train users and administrators to treat content processed by AI agents as potentially adversarial and to require independent validation before approving sensitive actions proposed as a result of externally sourced content.

References

Frequently Asked Questions

What is T1698 (Indirect Prompt Injection)?

T1698 is a MITRE ATT&CK technique named 'Indirect Prompt Injection'. It belongs to the Execution tactic(s). Adversaries may deliver malicious prompts to an LLM indirectly by embedding the prompts in content that the adversary expects may be processed by an LLM.(Citation: NIST indirect prompt injection) Thes...

How can T1698 be detected?

Detection of T1698 (Indirect Prompt Injection) typically involves monitoring system logs, network traffic, and endpoint telemetry. Use SIEM rules, EDR solutions, and behavioral analytics to identify suspicious activity associated with this technique.

What mitigations exist for T1698?

There are 6 documented mitigations for T1698. Key mitigations include: Restrict File and Directory Permissions, Privileged Account Management, Execution Prevention, Network Segmentation, User Account Management.

Which threat groups use T1698?

While specific threat group attribution may vary, this technique has been observed in various real-world attacks. Check the MITRE ATT&CK website for the latest threat intelligence.