Infographic showing an LLM' s knowledge cutoff.
Written by Martin Ågren - May 28, 2026
An LLM (a type of AI) has read texts up to a certain date.
This date is called the "knowledge cutoff".
The AI does NOT know anything that happened after that date.
This can lead to the AI providing incorrect information in its response.
Always ask yourself: "Is it reasonable that AI has read about this?"
Example, solution: Make sure AI searches online for the answer.
Sometimes it is smart to inform AI about today's date.
Each AI model is "created" after completing its training at a specific point in time, for example by reading almost all web pages on the internet, along with other texts.
Suppose the AI model finished reading all those web pages one year ago. In that case, that particular AI knows nothing about events taking place the past year.
Example, solution:
If you want to discuss something that happened recently:
Inform the AI of today's date (or ask the AI to find today's date online).
Ask your question and instruct AI to search online to find the answer.
As a manager or leader, it is important to understand that AI does not always have up-to-date knowledge so that this challenge can be handled appropriately.
Ensure that employees understand what the knowledge cutoff is and recognize the importance of enabling web search or providing LLMs with adequate information when prompting (for example, by pasting text or uploading a file).
Consider establishing simple guidelines in your AI policy for how knowledge cutoffs should be handled. This reduces the risk of AI-generated misinformation and improves the quality of AI use across the organization.