Start from what you observe.
Each page below starts from a symptom, not from a service: what you see in an AI answer, how to check that it is real, what can be corrected, and what no one can guarantee.
These pages are diagnostic. Each follows the same eight questions: how to recognize the problem, why it can appear, how to verify it, what can be corrected, what cannot be guaranteed, which evidence and sources are needed, how to measure improvement, and when an assessment is worth requesting. None of these situations is a crisis by itself.
What the answer says is wrong
- Inaccurate or outdated AI answers An assistant states a former role, an old description or a wrong fact as current.
- Confused with someone else The answer mixes you with a person or an organization of the same name.
You are not in the answer
- Not mentioned by ChatGPT An assistant has little or nothing to say about you.
- Competitors recommended by AI For a category question, the answer names others and leaves you out.
Only you describe you
- Weak independent sources Everything an assistant can find about you was written by you.
Before reading further
Check first that the problem is real. Ask the same question several times, in each assistant, and keep the answers: one run is a snapshot. The protocol is described in how AI visibility is tested. These five pages come from the six situations listed in what goes wrong; the sixth, diverging accounts, is treated in inaccurate or outdated AI answers. When the problem is confirmed and its cause is not clear, a private assessment is the starting point.
Understand how search and AI currently interpret your authority.
Request a confidential assessment of your visibility, entity signals, narrative consistency and source authority.
Request a Private AssessmentSelective engagements for leaders and organizations with material reputational stakes.