In a startling incident highlighting the dangers of artificial intelligence, ChatGPT, a popular AI chatbot developed by OpenAI, falsely accused George Washington University law professor Jonathan Turley of sexual harassment. The baseless claim, which cited a nonexistent Washington Post article, has raised serious concerns about the reliability of AI-generated information and its potential to spread harmful misinformation.
The incident came to light when UCLA law professor Eugene Volokh, conducting research on AI accuracy, prompted ChatGPT to provide examples of sexual harassment by U.S. law professors, including citations from news articles. Among the responses, ChatGPT alleged that Turley had made sexually suggestive comments and attempted to touch a student during a class trip to Alaska in 2018, referencing a supposed Washington Post story. However, Turley has never taught at Georgetown University as the chatbot claimed, never taken students to Alaska, and no such article exists. “It was quite chilling,” Turley told The Washington Post. “An allegation of this kind is incredibly harmful.”
Turley, a well-known constitutional law scholar and Fox News contributor, initially found the accusation absurd but soon recognized its “menacing” implications. In a USA Today op-ed, he wrote, “I have never been accused of sexual harassment or assault by anyone,” emphasizing that he has never taken students on any trip in his 35-year teaching career. The fabricated claim, complete with a fake citation, underscores what experts call AI “hallucinations”—instances where chatbots generate convincing but entirely false information.
This case is not isolated. ChatGPT also falsely accused an Australian mayor, Brian Hood, of serving prison time for bribery, prompting him to threaten a defamation lawsuit against OpenAI. Similarly, Microsoft’s Bing chatbot repeated the false claims about Turley, showing how quickly AI-generated misinformation can spread. Kate Crawford, a professor at USC and Microsoft researcher, described such errors as “hallucinations,” noting that chatbots, trained on vast internet data, often produce plausible-sounding falsehoods without fact-checking mechanisms.
The incident raises legal questions about accountability. Can OpenAI be held liable for ChatGPT’s defamatory outputs? Legal experts are uncertain. Section 230 of the Communications Decency Act typically protects online platforms from liability for third-party content, but it’s unclear if this applies to AI-generated text. Defamation claims require proof of harm, such as reputational damage, which Turley argues is a real risk given the convincing nature of AI responses. UCLA’s Volokh, who is studying AI-driven defamation, suggested that companies might face liability if they fail to correct false outputs after being notified.
Turley warned that AI’s ability to “metastasize” false claims across the internet poses a growing threat, especially as chatbots become more integrated into daily life. “AI algorithms are no less biased and flawed than the people who program them,” he wrote, pointing to the risk of confirmation bias in AI systems. OpenAI acknowledged the issue, stating, “We strive to be transparent that [ChatGPT] may not always generate accurate answers,” but critics argue more robust safeguards are needed.
As AI tools like ChatGPT gain popularity, this incident serves as a stark reminder of their potential to cause real-world harm. With no clear legal framework to address AI-driven defamation, experts and scholars like Turley are calling for greater scrutiny of these technologies to prevent reputations from being unjustly tarnished.
Reference from – ChatGPT falsely accuses law professor of sexual assault
