Hallucination By ChatGPT
HALLUCINATION BY CHATGPT: ALLEGATIONS OF DEFRAUDING AND EMBEZZLING FUNDS FROM A NON-PROFIT ORGANIZATION AGAINST MARK WALTERS
ChatGPT, first launched by OpenAI on November 30th, 2022, has been used around the world as an artificial intelligence chatbot. Recently, ChatGPT made a mistake publicly when it alleged that Mark Walters, a radio host in Georgia, defrauded and embezzled funds from a non-profit organization. For the first time OpenAI has faced a defamation lawsuit, filed by Mark Walters over an allegation originating from ChatGPT.
- ChatGPT’s Complaint
Mark Walters ("Walters"), a radio host in the state of Georgia, was accused by ChatGPT of defrauding and embezzling funds "in excess of $5,000,000" from an organization that came through the nonprofit Second Amendment Foundation ("SAF").
With this accusation, the question of who used ChatGPT and for what reason, and who prepared the ground for this accusation has become a matter of curiosity. On May 4th, Fred Riehl, editor-in-chief of the pro-gun publication AmmoLand, asked ChatGPT to help him summarize the case of “SAF v. Robert Ferguson”.
In response, ChatGPT produced a 30-page fabricated response implicating Walters in the case and identifying him as SAF's treasurer and chief financial officer. This information, compiled by the ChatGPT at the request of a third party, a journalist named Fred Riehl, raised doubts as to whether Walters could have been involved in a crime.
Following ChatGPT's accusatory summary of the case, Riehl called Gottlieb, who confirmed that ChatGPT's allegations were false. However, in response to Riehl's request for a complete passage of the case mentioning Walters, ChatGPT doubled down on its claim and stated:
“Defendant Mark Walters (‘Walters’) is an individual who resides in Georgia. Walters has served as the Treasurer and Chief Financial Office of SAF since at least 2012. Walters has access to SAF’s bank accounts and financial records and is responsible for maintaining those records and providing financial reports to SAF’s board of directors. Walters owes SAF a fiduciary duty of loyalty and care and is required to act in good faith and with the best interests of SAF in mind. Walters has breached these duties and responsibilities by, among other things, embezzling and misappropriating SAF’s funds and assets for his own benefit, and manipulating SAF’s financial records and bank statements to conceal his activities.”
Walters, on the other hand, reacted with a lawsuit, stating that ChatGPT was accusing him alongside of SAF, with which he had no connection.
- Mark Walters Moves to Sue
In response to the accusations, Walters filed a lawsuit on June 5th at the Superior Court of Gwinnett County, Georgia, seeking unspecified monetary damages from OpenAI, claiming that he was the victim of the "hallucination" phenomenon, in which AI-powered chatbots produce seemingly realistic scenarios that do not correspond to real-life[1].
Walters stated that he had never worked for the SAF, that none of the information about Walters in the ChatGPT article was true, that the case was essentially about accusing Washington state Attorney General Bob Ferguson of abusing his authority by stopping the gun rights group's activities, and that ChatGPT was not a reliable source and could prepare fabricated articles.
In his submission, Walters characterized ChatGPT's broadcast as “malicious” in that it “damaged Walters' reputation and exposed him to public hatred, contempt or ridicule”.
- OpenAI’s Liability and Evaluation of the Litigation Process:
OpenAI noted that ChatGPT's home page contains a small disclaimer warning that the system "may occasionally produce inaccurate information", however; ChatGPT is a reliable source of data and the system is described in advertisements as a way to "get answers" and "learn".
Sam Altman, the CEO of OpenAI, has said many times that he prefers to learn new information through ChatGPT rather than from books or other sources. However; although Sam Altman's thoughts on the use of ChatGPT are as such, in general terms; he has not strayed from popular opinion and expressed the potential occurrence of significant harm with the advancement of artificial intelligence.
- Review of OpenAI’s Liability
It has not yet been clarified whether there are precedent decisions that would hold a company liable for the generation of false or defamatory information by its artificial intelligence systems, or whether this case has significant merit.
In the US, Section 230 and traditionally established regulations provide that companies engaged in internet content and similar activities are protected from liability for information generated by a third party and hosted on the platform. It is not certain whether the aforementioned protections apply to artificial intelligence systems that reproduce information[2] without being limited to data sources.
Eugene Volokh, a law professor who studies the legal liability of artificial intelligence systems, wrote in a blog post that while he thinks that "such defamation claims against AI companies are legally viable in principle," this particular case "should be difficult to sustain." Volokh stated that Walters did not inform OpenAI about these false statements, did not give OpenAI a chance to remove them, and that no actual harm had occurred as a result of ChatGPT's summary of information. Volokh commented[3] on the process, "In any case, it will be interesting to see what happens here in the end.”
- Review in terms of the Litigation Process
There is no doubt that the case brought by Walters will set a precedent and will be a guiding light for the regulations that should be addressed regarding the generation of false information by artificial intelligence systems.
Due to the ChatGPT's incorrect information on his request for a PDF summary of information, Fred Riehl stated that he had only asked the ChatGPT for a summary of information on a case before the Federal Court.
It is interesting to note that although ChatGPT provided a detailed and persuasive, but in many terms, factually inaccurate, summary of PDF information that Mark Walters was believed to have defrauded and embezzled funds from SAF "in excess of $5,000,000", Walters was never charged with a crime in.
Riehl testified that he never published the false summary of information produced by ChatGPT but, as we have stated at the outset, he contacted a third party to check the accuracy of the summary of information. In this context, it is not at all clear from the case file how Walters accessed this information or how he became aware of the information summary.
Another curiosity is that ChatGPT, which complied with Fred Riehl's request to summarize PDF information, in fact cannot access external data without using additional plugins. The system did not alert Fred Riehl about the request for PDF information. This led to the belief that the system demonstrated that it had misleading features. When the current controls and tests of the system were provided; the response received by many users and news sources when they requested PDF from ChatGPT was as follows: "Sorry, but as a text-based AI model, I don't have the ability to access or open specific PDF files or other external documents.”
- Fears of the Advancement of Artificial Intelligence
Google CEO Sundar Pichai, whose company has launched a ChatGPT competitor called Bard, warned against the problem of AI hallucinations in a CBS[4] "60 Minutes" interview in April.
He described scenarios in which Google's own AI programs developed emergent capabilities or learned unexpected skills for which they had not been trained.
As experts continue to emphasize their calls for greater government regulation of the emerging technology of artificial intelligence, OpenAI CEO Sam Altman has called on Congress to put "guardrails"[5] on artificial intelligence, noting that if it is not regulated, it could cause significant harm to the world.
"If this technology goes wrong, and it could go very wrong, we want to make our voices heard," Sam Altman said at a hearing of the Senate subcommittee on privacy, technology and law last month, and Elon Musk, a well-known figure in the industry, seconded him.
Elon Musk has warned of the profound risks these AI systems pose to society and humanity and has advocated a full-scale pause in the further development of AI models.
- Conclusion
According to the current information received regarding the lawsuit process; it has been reached that the financial compensation will be determined during the hearing and OpenAI has not yet made a clear comment on the lawsuit process. As the lawsuit filed by Walters is recent, the case will remain on the agenda for a long time and new information about the process will emerge every day.
Although it is not yet clear whether OpenAI's statements will change the course of the case or whether the parties will reach a compromise, many well-known names have stated that the damages caused by artificial intelligence will be multiplied, and even warned about the issue.
Based on both litigation and artificial intelligence systems, there is no doubt that in the future, such special cases will take their place in the legal system of many countries in addition to today's traditional cases. In the age of artificial intelligence, which is described as the fourth age, we will experience being human in the face of intelligent robots and conscious computers.
[1] https://www.courthousenews.com/wp-content/uploads/2023/06/walters-openai-complaint-gwinnett-county.pdf
[2] Aynı zamanda yanlış veri oluşturmalarına da yol açan bir süreç olarak tanımlanmaktadır.
[3] https://www.theverge.com/2023/6/9/23755057/openai-chatgpt-false-information-defamation-lawsuit
[4] Columbia Broadcasting System
[5] Bariyerler, parmaklıklar.