AI Tools Are Linking to Articles and Sources that Don’t Exist

Sometimes, when you ask your chatbot questions they will cite their sources to you along their answer. But how often do you check them out?

At the Danish science and news site Videnskab.dk, they noticed traffic coming from AI tools like ChatGPT, Microsoft Copilot, Claude, and Perplexity.

People click on links that look like our articles. Some are about gender, mental health, or Greenlandic children.

But here’s the catch: many of those articles don’t exist.

The URLs look real. But they lead nowhere. The AI made them up.

According to AI expert Peder Hammerskov, this is a common problem. He teaches at the Danish School of Media and Journalism.

He explains that AI tools often guess what a source could look like. They aren’t trying to lie.

They just want to give us a helpful answer. The result is sometimes fiction.

This is called a “hallucination.” The AI fills in gaps when it lacks information. It knows how articles usually look.

It knows how we write and what we expect. So it makes something up that seems believable.

That can lead to fake links, made-up quotes, or even imaginary studies.

In one case, a journalist asked an AI to write a story based on a restaurant report. The AI invented a source and quotes.

It looked good but wasn’t real. And sometimes, these hallucinations make it into real news.

Just last week, a U.S. newspaper published a summer reading list. It had fake book titles created by AI.

So how do you protect yourself? Hammerskov suggests checking every link. See if the source exists.

Ask if it’s reliable. Some tools have research features, but you still need to double-check. Don’t treat AI as a fact machine. It’s more like a conversation partner.

His advice? Slow down. Take time to verify. If the AI gives no sources, go find your own.

That’s how you make sure the information is actually true.