AI assistants describe companies with great confidence, even when the details are out of date or wrong. An old head office, a discontinued product line or a former managing director can end up in answers that potential customers take at face value. You cannot prevent this entirely, but you can influence it.
Where do incorrect details in AI answers come from?
ChatGPT, Gemini and Perplexity draw on two things: the knowledge the model was trained on, and web pages they retrieve when a question is asked. Errors usually arise in one of three ways. The training knowledge is out of date because it has a cut-off date. The web contains conflicting information, for example an old entry in a business directory next to the current website. Or there are too few good sources, and the model fills the gap with a plausible-sounding guess.
How do I find out what AI says about my company?
Ask the questions your customers would ask. Not just “What is company X?”, but also “Which providers of Y are there in Vienna?” or “Who is the managing director of X?”. Repeat this in ChatGPT, Gemini and Perplexity and note which details are correct, which are missing and which sources are cited. The sources usually show where the error comes from.
What helps to correct wrong information?
The most important step is to publish the correct information where AI systems actually look. That starts with your own website. An “About us” page that states the founding year, locations, management and services in plain sentences is more valuable to search systems than a polished image film. Structured data according to Schema.org also helps to attribute names, addresses and people correctly.
Next come the sources the AI cites for your company. Outdated entries in business directories, on review platforms or with chambers and associations should be updated. If there is a Wikipedia article, take a look. Rather than editing it yourself, request a correction on the talk page and provide evidence.
Current media coverage has the greatest effect. A new article in a trade publication that names the new managing director or the new product line is picked up quickly by AI systems with web search. A press release about an appointment, a move or a change of direction is therefore also a tool against outdated AI answers.
How long does it take for answers to change?
Systems with web search, such as Perplexity or ChatGPT with search enabled, often adopt new information within days or a few weeks once the new sources can be found. The stored knowledge of the models only changes with the next training run, so much later. That is why it is worth checking the details regularly, for example once a quarter.
What should you avoid?
Some providers promise to “overwrite” AI answers with large volumes of generated content. We advise against this. Search systems increasingly filter out such pages, and they can damage your reputation. What works in the long run is what people also find credible: a well-maintained website, correct listings and genuine media coverage.
GEO at Fullstop PR
We check what ChatGPT, Gemini and Perplexity say about your company and make sure it is correct.
Frequently asked questions
Can I ask OpenAI or Google to remove incorrect information?
The major providers offer forms for correction or deletion requests concerning personal data. For company information, the more effective route is to correct the sources online and publish up-to-date information.
Does a press release help against incorrect AI answers?
Yes, if it states the correct facts clearly and is published openly, for example in your own newsroom and via a press distribution service. The coverage it generates in trade media has an even stronger effect.
How often should I check what AI says about my company?
Once a quarter, and additionally after major changes such as a change of leadership, a move or a rebrand.
