What is a knowledge cutoff?
A knowledge cutoff is the date after which a language model has seen no training data. Anything newer, such as a product launch, a price change or a rebrand, is unknown to the model unless it searches the web. Providers publish the cutoff for each model, and it is usually months before release.
Consequences for your brand
A model without search can describe an old version of your company. Web search fixes part of this, but the model may still prefer what it remembers when sources conflict.
Example: a model with a cutoff in spring does not know about a rebrand in autumn unless it searches the web.
What to do
Update every public profile when facts change, publish the change on a dated page, and re-ask the key questions after a few weeks.
Check it for your brand
Nodpick re-asks your questions every week, so you see when an assistant starts to know about your latest change. Run the free AI visibility check to see your own numbers; the method shows how they are made, and pricing lists the plans.
Read more: How ChatGPT chooses which brands to recommend