ChatGPT defames a mayor
The mayor of Hepburn Shire, an Australian city northwest of Melbourne, has threatened to sue OpenAI for defamation. The reason? For some strange reason, ChatGPT claims that the man has been in prison for corruption. The news is particularly interesting as it could represent the first defamation lawsuit ever brought against an AI chatbot.
Brian Hood, this is the name of the mayor, would have discovered the fact when informed that, according to ChatGPT, he would have been implicated in a corruption scandal involving Note Printing Australia (a company of which Hood was CEO) concerning paid bribes to officials in Malaysia, Indonesia and Vietnam.
Credit: Artisana
--> The Artisana.ai site questioned ChatGPT on the issue, obtaining the answers you see in the screenshots below. The chatbot falsely claimed that Brian Hood received a sentence of four years and six months after pleading guilty to a charge of conspiracy to bribe a foreign public official. According to ChatGPT Hood would have played a central role in the corruption plan, which is why he would also have been fined 130,000 Australian dollars. When the AI was asked for sources, ChatGPT responded with three links to stories from ABC News, The Guardian and Reuters; even if the titles of the articles and the links seemed plausible and correct, in reality the URLs were non-existent and did not lead to any web page.
However, the reality is very different from what was described by the AI: it was indeed Hood to report the bribes to the authorities, moreover he has never been charged with any crime. The lawyers representing him sent a letter to OpenAI on March 21, giving the firm 28 days to correct the error regarding their client, or face a defamation lawsuit. According to the lawyers, OpenAI has not yet responded to the request or commented.
If the Hood's lawsuit against OpenAI were launched, it would be the first case ever registered against an AI chatbot, undoubtedly a historic moment for the law on defamation. The lawyers also argue that reputation is paramount to the mayor, given his role as a public servant; having false information like this could be harmful.
--> However, it should be emphasized that chatbots like these are prone to generating "hallucinations" like this one, i.e. falsehoods that sound real. The BBC reported that different chatbots give different answers to the question: for example, Bing's Chat correctly identifies Hood as an informant, without falsely accusing him. The story undoubtedly raises important questions about the use of artificial intelligence and the responsibilities of the companies that develop them, but to avoid unpleasant situations, it is always a good idea to verify what the chatbots are saying by checking authoritative sources on the net.