AI Will Confidently Make Up Cases, Studies, and Articles That Don’t Exist

Here’s the thing about AI hallucinations: they don’t look like mistakes. They look like facts. The AI gives you a confident answer with a real-sounding case name, a plausible journal, a believable quote. And if you don’t check, you’d never know it made the whole thing up.

This isn’t rare. Stanford HAI found that AI legal tools hallucinate in at least 1 out of every 6 queries — and earlier research on general-purpose chatbots found they hallucinated between 58% and 82% of the time when asked legal questions. That’s not a small bug. That’s a massive, ongoing problem.

And it’s already costing real people. Two lawyers — Christopher Kachouroff and Jennifer DeMaster — were sanctioned by a federal judge after filing a legal document packed with more than two dozen errors, including fake cases that AI had completely invented. These weren’t students submitting homework. These were practicing attorneys.

It’s not just law. Researchers published in PubMed Central actually tested this by asking ChatGPT to write a fake medical research article. It did. In hours. The result was a convincing, realistic-looking scientific paper — completely fabricated. Their conclusion? The fake article would still need expert review and real improvements before it could be submitted to a journal, but the fact that AI got that far is alarming.

So what does this mean for you? If you’re using AI to research anything that actually matters — a school paper, a health question, a legal situation, a news story — you can’t just trust the output. You have to verify. Look up every source the AI gives you. Check that the case, study, or article actually exists. If you can’t find it through a real database or library search, it might not be real.

AI is a powerful tool. But it’s a tool that will lie to you with a straight face and zero hesitation. Knowing that changes how you use it.