Find out what AI assistants tell people about you
What ChatGPT, Gemini, Claude, Perplexity, Copilot and Google AI Overviews say about you, and why.
An AI Reputation Audit tests a structured set of questions about you or your organisation across the major AI assistants and AI search features, records every answer, checks it for accuracy and tone, and traces the sources each assistant relies on. That source map is what makes correction possible.
Who this is for
- Businesses whose customers now research through AI assistants
- Executives and public figures concerned about inaccurate AI summaries
- Organisations recovering from a past issue that AI answers still repeat
Who it is not for
- Anyone expecting a guarantee of specific wording from an AI system. No one can promise that.
What the work involves
Prompt set design
Questions a real customer, journalist, investor or employer would ask, in several phrasings.
Multi-assistant testing
Answers captured from each assistant, with and without web search where the product allows.
Accuracy and sentiment grading
Every claim is checked against the record and graded accurate, partly accurate, inaccurate or missing.
Source tracing
The pages each assistant cites or appears to draw on, ranked by influence.
Correction plan
Which sources to correct, strengthen or create so future answers improve.
How the work runs
- Confidential intakeNDA signed before you share details. We confirm scope, subjects and jurisdictions.
- Evidence captureWe record search results, reviews, media, AI answers and data exposure with dated screenshots.
- AnalysisEach item is classed by harm, reach, accuracy and the legal or policy routes available.
- Findings sessionYou receive the written report and a written findings review with your case lead through the Confidential Consultation System.
What you receive
- Answer log with dated captures
- Accuracy and sentiment scorecard by assistant
- Source influence map
- Correction plan
What we cannot promise
- AI outputs vary between sessions and change without notice. We measure patterns across repeated tests, not single answers.
Frequently asked questions
Can you change what ChatGPT says about me?
Not directly. Assistants draw on the sources they can find and trust. We improve and correct those sources, which in our experience is what moves the answers over time.
Which assistants do you test?
ChatGPT, Gemini, Claude, Perplexity, Microsoft Copilot and Google AI Overviews as standard. Others can be added.
How often should this be repeated?
Quarterly for most clients, or monthly as part of a monitoring retainer.
What if an AI answer is defamatory?
We document it and, where appropriate, coordinate with a solicitor and use the provider's reporting routes.
Do you use my data to train anything?
No. Prompts and answers are stored only in your case file.
Is my enquiry confidential?
Yes. You can make first contact without giving your name, we sign a non-disclosure agreement before you share case details, and case files are held in restricted, encrypted storage with access limited to your case team.
Prefer to start yourself?
Removal guideHow can I change what ChatGPT says about me?
Read the guideRelated services
Often combined with
AI Search Reputation Correction
Correct inaccurate AI answers by fixing the sources assistants rely on.
Learn more RepairKnowledge Panel and Wikidata
Accurate knowledge panels and entity data, within Google's and Wikipedia's rules.
Learn more ProtectReputation Monitoring and Intelligence
Search, reviews, news, forums, social and AI answers, watched on a set cadence with graded alerts.
Learn moreStart with a confidential conversation
Tell us as much or as little as you like. We sign an NDA before you share case details, and we will tell you honestly whether we can help.