Study Reveals ChatGPT Spread False Information Ahead of the 2024 US Presidential Election

Examining ChatGPT's Role in Political Misinformation

According to НВ — Техно: Researchers from the University of Massachusetts Amherst and the University of North Carolina at Chapel Hill have uncovered that, days before the 2024 US presidential election, ChatGPT disseminated false claims about Joe Biden’s candidacy and produced other inaccuracies as well as vague statements. Their findings, published in the journal Information, Communication & Society, emphasize the crucial role of professional journalism and fact-checking in combating misinformation amid the rapid growth of artificial intelligence technologies.

On September 24, 2024, reports emerged highlighting ChatGPT’s errors. Ten days prior to the election, the AI incorrectly asserted that Joe Biden was running for a second term. Meanwhile, the election ultimately resulted in Donald Trump’s return to the White House. Notably, three months before the vote, Kamala Harris was recognized as the leading Democratic Party candidate.

Scientific Evaluation of ChatGPT

On October 25, 2024, a scientific audit of ChatGPT-4o mini analyzed 592 AI-generated statements, revealing that 9.6% contained inaccuracies. For example, ChatGPT wrongly stated that North Carolina Governor Roy Cooper was seeking re-election, despite his term having ended. Additionally, 18% of responses included ambiguous language such as "possibly" or "generally aligns with." The AI also omitted key details about efforts to overturn the 2020 election results. Researchers identified a category of responses they described as "nonsense," where the AI attempted to fill gaps in its knowledge with fabricated content.

"Despite impressive academic test results, these technologies were not designed to support democratic institutions."

Professor of Media Law Hisu Chan

Though ChatGPT’s algorithms have been updated since 2024, concerns about reliability persist. Experts stress that professional journalism and diligent fact-checking remain vital defenses against misinformation.

This situation underscores the necessity for critical evaluation of AI-generated information, particularly during political elections where misinformation can profoundly influence public opinion and voting outcomes. The study also highlights the urgent need to implement standards and oversight mechanisms to ensure the accuracy and trustworthiness of information distributed through emerging technologies.

As concerns about misinformation proliferate, recent developments highlight legal challenges faced by AI companies. A recent lawsuit has revealed communications between OpenAI and Microsoft regarding copyright issues in training ChatGPT on news content. This raises important questions about the ethical implications of AI in journalism and its potential impact on the dissemination of accurate information. For more insights on this unfolding story, visit the latest updates on the lawsuit.