Of course, as with all other information that an AI spits out, you can't trust it 100%. But it is a very good indication of whether the information is publicly available at all or whether you should contact support immediately, for example. For example, I like to ask the AI about the sources used and then take a look at them. Grok even goes one step further here and displays the sources used directly:



But basically you're absolutely right: don't trust, check