Skip to content

MITRE ATT&CK technique

T1698 — Indirect Prompt Injection

Last fetched

T1698 (Indirect Prompt Injection) is a MITRE ATT&CK technique tracked in WhisperGraph, serving the Execution tactic.

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](https://attack.mitre.org/techniques/T1685/007), and for unauthorized execution of any capabilities available to the LLM.

Tactics

Threat actors observed using this technique

No threat actor is recorded using T1698 in WhisperGraph.

© The MITRE Corporation. This work is reproduced and distributed with the permission of The MITRE Corporation.

Related pages

Pivot from T1698 into its tactic, related techniques and the actors that use it.

Queries

Resolves the segment to this technique or tactic.


MATCH (t:ATTACK_PATTERN {name: $id})
RETURN t.id AS id, t.name AS name, t.kind AS kind, t.description AS description
Run yourself →

The tactic(s) this technique serves.


MATCH (t:ATTACK_PATTERN {name: $id})-[:USES_TACTIC]->(tac:ATTACK_PATTERN)
RETURN tac.id AS id, tac.name AS name
LIMIT 10
Run yourself →

This id's parent technique, if it is a sub-technique.


MATCH (p:ATTACK_PATTERN {name: $parentId})
RETURN p.id AS id, p.name AS name
LIMIT 1
Run yourself →

Sub-techniques of this technique, if any.


MATCH (c:ATTACK_PATTERN) WHERE c.id STARTS WITH $subPrefix
RETURN c.id AS id, c.name AS name
ORDER BY c.id
LIMIT 25
Run yourself →

Threat actors observed using this technique.


MATCH (a:ACTOR)-[:USES_TECHNIQUE]->(t:ATTACK_PATTERN {name: $id})
RETURN a.name AS name
LIMIT 100
Run yourself →

Or query Whisper from your own LLM workflow via the Whisper MCP server.