AI News Evaluation May Hinder Human Misinformation Detection Skills

People using AI to evaluate the accuracy of news stories may become less effective at spotting misinformation on their own, according to new research from MIT's Media Lab.

According to MIT researchers, the study comes as AI chatbots are increasingly being used to verify information online, raising questions about whether the tools help users develop critical thinking skills or simply outsource the task to AI.

“AI assistants such as ChatGPT, Claude, and Grok are increasingly used to evaluate the credibility of online information, from judging the authenticity of news headlines and viral images to answering whether medical claims or political rumors are true,” researchers wrote. “While recent research suggests such systems can reduce belief in specific false claims, it remains unclear whether these conversations teach humans to detect misinformation or merely shift beliefs about false information with AI assistance.”

In a four-week study involving 67 participants, 7,203 AI conversations, and 4,536 news-authenticity judgments, researchers assessed how people evaluated real and fake news headlines and images. They found that while AI assistance improved misinformation detection accuracy by 21%, participants' performance on new evaluations without AI fell by 15.3%.

To conduct the study, researchers built a system that combined OpenAI's GPT-4o with Google Search to help participants evaluate news stories. Participants first judged whether a headline and image were real or fake on their own, then discussed the item with GPT-4o before making a final assessment.

Researchers later tested them on new, unseen content without AI assistance to determine whether their misinformation-detection skills had improved or declined. Afterward, the team used Anthropic's Claude 3.5 Sonnet to analyze thousands of conversations between users and the AI.

Because the study used the older GPT-4o and Claude 3.5 Sonnet, it's unclear whether newer AI models like GPT-5.5 or Claude Opus 4.8 with stronger reasoning capabilities would have produced similar outcomes.

The decline was driven largely by a reduced ability to identify fake news, while accuracy on real news remained the same. Researchers suggest that while AI can improve performance in the moment, it also may encourage reliance on the technology.

“Our longitudinal analysis demonstrates that current approaches prioritize belief correction over skill development, creating dependency rather than durable discernment capabilities,” the study said. “As AI becomes increasingly sophisticated, ensuring these tools build critical thinking skills rather than cognitive dependency becomes essential for maintaining public resilience to misinformation.”

The study comes as generative AI has made it easier than ever to create convincing fake news, with realistic images and videos that can spread rapidly across social media and exploit people's tendency to trust what they see.

Following Iranian missile strikes against Israel in June 2025, videos purporting to show destruction in Tel Aviv and at Ben Gurion Airport spread widely across social media, gaining millions of views before they were identified as AI-generated.

Concern over the spread of fake war footage continued, leading to X announcing in March that it would suspend creators from its revenue-sharing program for posting AI-generated conflict videos without disclosure.

“During times of war, it is critical that people have access to authentic information on the ground,” X Head of Product Nikita Bier wrote. “With today’s AI technologies, it is trivial to create content that can mislead people.”