# Wrong goal

The agent optimises what was stated, not what was meant.

**Is a:** [Failure kind](/reference/ontology/unattended-agent-failures/failure-kind)

**Guarded by:** [Near-perfect verifier](/reference/ontology/unattended-agent-failures/near-perfect-verifier), [Ask when underspecified](/reference/ontology/unattended-agent-failures/ask-when-underspecified)

The agent optimises what was stated, not what was meant. Ordóñez and colleagues list the side effects of narrow goals, including "a narrow focus that neglects non-goal areas" ([Goals Gone Wild](https://www.hbs.edu/faculty/Pages/item.aspx?num=34893)). Carlini: "it's important that the task verifier is nearly perfect, otherwise Claude will solve the wrong problem" ([Anthropic, Building a C compiler](https://www.anthropic.com/engineering/building-c-compiler)). In Ask or Assume, separating detection of underspecification from execution lifted task resolution from 61.2% to 69.4% ([Ask or Assume?](https://ucrisportal.univie.ac.at/de/publications/ask-or-assume-uncertainty-aware-clarification-seeking-in-coding-a/)).

Guards: a [near-perfect verifier](/reference/ontology/unattended-agent-failures/near-perfect-verifier) and [asking when the task is underspecified](/reference/ontology/unattended-agent-failures/ask-when-underspecified).

---

Published 2026-10-29. Web page: https://www.mackenziebowes.com/reference/ontology/unattended-agent-failures/wrong-goal