Artificial intelligence is supposed to make our lives easier, but it’s also becoming a vector for misinformation — and that’s raising serious concerns. AI chatbots are failing to catch false claims, sometimes spreading them instead.
AI Models Fail to Catch False Claims
Recent tests show AI models like ChatGPT, Grok, and Gemini are not only missing false content but also reinforcing it. One model insisted an AI-generated image of a Ryanair flight was real — when it wasn’t.
- Another claimed an image from the Artemis II mission was authentic — when it wasn’t
- Models were asked about content Full Fact was already investigating — so they weren’t just failing to catch misinformation
- They were feeding false information back to users
Misinformation Spreads Beyond Images
AI chatbots are also misidentifying political banners, fake council posters, and videos. One model said a fire in Saudi Arabia was actually a missile strike — which is alarming when false narratives spread quickly.
AI’s Role in Other Scandals
This isn’t the first time AI has been linked to misinformation. In 2022, a voice actor found out his voice was cloned — without permission — and sold as an NFT. That scandal sparked debates about AI ethics, ownership, and abuse.
New Cybersecurity Threats Emerge
Researchers at the University of Montana-Missoula are warning that generative AI is creating new cybersecurity threats — from deepfakes to phishing attacks. They stress the need for users to be vigilant, especially as AI tools become more sophisticated.
AI’s Growing Influence on Misinformation
Generative AI is becoming a key player in spreading false information — and governments are struggling to keep up. Australia’s proposed laws on mis- and disinformation have been criticized for lacking clear definitions — especially when it comes to AI’s role.
What Can Be Done?
Experts are calling for better transparency, stronger regulation, and more user education. But with AI models getting smarter — and malicious actors getting more creative — the challenge is only going to get harder.
Can You Trust AI?
You might be wondering — if AI can’t tell the difference between real and fake content, how do you trust it at all? It’s a fair question — especially when so many of these models are used in newsrooms, political campaigns, and law enforcement.
AI isn’t the enemy — but it’s also not a perfect solution. As you rely more on these tools, be clear-eyed about their limitations. Because if you don’t — the next time a chatbot tells you something false, it might be too late to catch it.
