Knowledge cutoff
Definition
A knowledge cutoff is the date after which an AI model has no information from its training. Anything that happened later, such as news, price changes or new product releases, is unknown to the model unless it searches the web or reads a document you give it.
By Ronan Liedmeier, fractional head of growth and course author · Published 17 October 2026 · Last updated 17 October 2026
Also called: training cutoff, training data cutoff
Models learn from a fixed snapshot of text and stay in use for months or years after it was taken. Ask about this quarter's prices or last week's announcement and a model without search may answer anyway, from older data, with no warning.
How it works
The cutoff is fixed when the training data is collected. Providers publish it in their model documentation; Anthropic, for example, lists a "reliable knowledge cutoff" for each Claude model.
Web search and connected tools are how assistants get past it. When a chat assistant searches, it reads current pages and answers from them. When it doesn't, it answers from memory, and that answer can be months or years out of date. An outdated answer stated with confidence is a form of hallucination.
Example
A recruiter asks an assistant for the current minimum salary threshold for a skilled-worker visa. The first reply gives a figure from memory with no source. She asks it to search the official government site and cite the page. The number changes, because the threshold was raised after the model's training data was collected.
Common mistake
People take an answer as current because it sounds current. Many assistants decide on their own whether to search, so look for live sources in the reply. For prices, laws, interest rates and recent events, ask for a web search with links, or paste in the current document yourself.
Learn it in AI Academy
- Fact-Checking & Spotting HallucinationsBeginner · 10 min
- How AI Actually Works (The 20% You Need)Beginner · 12 minFree
Related terms
- Hallucination
- A hallucination is a false statement an AI model presents as fact, such as a made-up statistic, quote or source. It happens because language models are trained to produce likely-sounding text, and likely-sounding text is sometimes wrong.
- Deep research
- Deep research is a mode in AI assistants where the model runs many web searches on its own, reads the sources it finds and writes a longer report with citations. It takes longer than a normal answer and suits questions that need several sources compared.
- Large language model (LLM)
- A large language model (LLM) is an AI system trained on huge amounts of text to predict which words should come next. At scale, that one skill lets it answer questions, summarize, translate and write in a way that reads like a person wrote it.