Question and Answer

How can I fact-check the information that ChatGPT and other language models give me?

If you are using a model that links to its sources (like Copilot, Perplexity, or Gemini), follow the links and read the original pages. Make sure the AI-generated summary aligns with the content of the page it came from. And make sure the page content is relevant to the task you asked the model to do.

If you are using the free version of ChatGPT (without links to sources), you will want to do a quick web search to find out if what it’s saying is true. Look for more than one source to verify the information. Wikipedia can be helpful as can mainstream news sites that employ fact-checkers.

Because the free version of ChatGPT doesn’t have an understanding of facts, it’s often better to use a model that links to its sources, like Perplexity or Copilot. This makes it easier to fact-check.

Since websites can also contain misinformation, try using the SIFT Method: Stop, Investigate the source, Find better coverage, and Trace claims to the original context.

Note: As of May 13, 2024, free accounts on ChatGPT will begin to get access to a more powerful version called GPT-4o. It’s being rolled out gradually over a few weeks to free accounts. It includes web browsing, so it can look up more current information. However, it has a limit to the amount of use in one day, so when you reach your limit, you’ll automatically be rolled back to the earlier version, which doesn’t include web browsing. You’ll have to wait until the next day to use the more powerful model again.

Learn more

Related FAQs

    Frequently Asked Questions