How do we control what ChatGPT says about us?

Direct control of ChatGPT output is not on the table for any company. The model is proprietary, the prompts are user-controlled, and asking the model to change its answer has no durable effect across sessions or users, ChatGPT does not learn from or remember individual conversations. What works is influencing the sources ChatGPT relies on and monitoring continuously so drift is caught early.

The levers that actually move the answer

AI reputation is a product of the sources the engines read, and that is where the work happens. In our experience the levers, in rough order of influence, are:

• Wikipedia, the single biggest lever. It is one of the most-visited sites on the internet and a reference AI engines lean on heavily.
• Mainstream news coverage, reputable outlets maintain corrections policies and will fix documented factual errors when they are properly sourced.
• Structured data, the machine-readable facts that feed knowledge graphs and entity records.
• Well-built owned properties, your own sites and profiles, kept accurate and crawlable.

Why “just tell the model” doesn’t work

The answer is built from the sources. Trying to manage the answer without managing the sources is theater: the engine doesn’t remember the conversation, so a correction typed into one chat session vanishes when the next user opens a fresh one, and the engine rebuilds the answer from the sources it trusts. Lasting change comes from those sources, not the chat box.

Continuous monitoring closes the loop

Because no fix is permanent and the models keep changing, the source work has to be paired with ongoing measurement. AIQ polls eight major AI engines: ChatGPT, Gemini, Copilot, Perplexity, Claude, Grok, Google AI Overviews, and Google AI Mode, so that when the narrative drifts from accurate or fair, it is caught early and the right source can be addressed.