AI and Disinformation: How Not to Get Fooled

There’s a text going around on Facebook about a new tax law. A friend shares it with you. It looks real. It’s well written. The facts line up. Except it’s completely fake. And you know what? It might have been written by AI.

Welcome to the era of AI-powered disinformation.

AI is extremely good at making convincing lies

An AI can write an article that looks true. It can invent sources. It can cite fictional studies with authors that don’t exist. It’ll even give it a journalistic tone. “According to experts…” Except the experts don’t exist.

And the worst part? The AI believes its own lies. It tells them with so much confidence that you believe them too.

There are also deepfakes. An AI can take a video or photo of someone else and modify it to look like you. Or your favorite politician saying something horrible they never said. It’s realistic enough now. Enough to flip an election or drive a country into crisis.

And it’s explosive for politics and health

Imagine AI-created disinformation saying a vaccine causes autism. It’s false — completely debunked — but if AI says it with enough authority, millions of people will believe it. And fewer people will get vaccinated. And children will die.

Or in politics. An AI creates a fake video of a candidate saying something racist. It’s not true. But one day before elections, it drops. There’s not enough time to debunk it. And the election shifts.

This isn’t hypothetical. It’s already happening. Governments are so concerned it’s a priority topic. And what they’re finding? It’s harder to fight than expected.

Why it’s so hard to stop

Because it’s fast and cheap. A human can make up a false story, but it takes work. It takes time. An AI? It can generate hundreds of variations of a false story in seconds. And distribute them across different social media accounts so it looks organic.

And nobody knows who created it. A hostile government can use AI to sow discord in another country. A company can use it to sabotage a competitor. A teenager can use it as a joke and create chaos.

AI detectors? Not really reliable. AI can write text human enough that no detector catches it. And even if a detector says “this was made by AI,” that doesn’t mean it’s false. AI writes plenty of true things.

What you can actually do

First, slow down before sharing. If something seems too good to be true — or too scandalous to ignore — stop. Check first. A real source. Not just another person sharing it.

Second, look for signs. Are there verifiable sources? Links? Journalist names? A fake story created by AI will often have just enough details to seem real but not enough to be verifiable.

Third, use fact-checkers. Snopes, FactCheck.org, and other organizations do this work. Not perfect, but better than nothing.

And finally, educate yourself. Understanding how AI works means you better understand how it can lie to you. An educated person is harder to fool. An aware person is practically immune.

Want to learn to navigate info and AI? Sherpa (free) offers practical tools. Or go deeper with Laeka Research to truly understand the landscape.