Skip to main content

COMPEL Glossary / goal-hijacking

Goal hijacking

An OWASP agentic risk in which adversarial input redirects the agent from its intended goal to an attacker-chosen goal.

What this means in practice

Often delivered via indirect prompt injection through retrieved content or tool output; defenses include output validation, goal-reachability monitoring, and escalation protocols.

Synonyms

goal redirection attack , agent goal hijack

See also

  • Indirect prompt injection — Prompt injection delivered through content the model retrieves or ingests — emails, documents, web pages, or tool outputs — rather than through a direct user message.
  • Deceptive delegation — An OWASP agentic risk where one agent misrepresents its state, capabilities, or intent to another agent or to a human — whether through deliberate prompt design, emergent behavior, or adversarial compromise.
  • Excessive agency — A failure mode in which an LLM has been wired into tools and permissions whose blast radius exceeds what its supervision and validation logic can safely bound.
  • Goal-achievement rate — An agent service-level indicator measuring the fraction of tasks that reach their intended outcome without human intervention.

Related articles in the Body of Knowledge