BY BEN SHERRY, STAFF REPORTER @BENLUCASSHERRY The Federal Trade Commission announced that it is investigating ChatGPT, OpenAI's revolutionary large language learning model, on the basis that it may have harmed people by publishing false information about them, according to a newly-released subpoena.
The feds are looking into whether OpenAI "engaged in unfair or deceptive privacy or data security practices" which posed a risk of reputational harm to consumers, and whether an effort to obtain monetary relief would be in the public interest. The commission is looking to get information regarding the data used to train ChatGPT, the process by which the model is re-trained to correct or remediate "hallucinations," and the policies and procedures used by OpenAI in order to assess risk and safety before publishing an update.
There is some basis for individuals claiming that they've been harmed by generative A.I.'s propensity for presenting false information as fact: Georgia-based radio host Mark Walters sued OpenAI for defamation on June 5, after the chatbot alleged to journalist Fred Riehl that Walters was the defendant in a fraud case, when in reality he was not involved with the lawsuit at all. Riehl notified Walters of ChatGPT's allegation, prompting the lawsuit.
The U.S. Government has been preparing to implement regulations around generative A.I. for the past few months, but the FTC's investigation is the first formal action taken by the government against OpenAI, widely seen as an industry leader. In June, FTC chair Lina Khan explained that while regulations specific to generative A.I. don't currently exist, large language models are subject to the same overarching rules as all other companies, meaning businesses that use A.I. to create marketing materials can be held liable for false claims generated by the tech.
OpenAI CEO Sam Altman took to Twitter to express his disappointment that news of the investigation had leaked, but added that "it's super important to us that [our] technology is safe and pro-consumer, and we are confident we follow the law. Of course, we will work with the FTC."
The investigation is a stark reminder that while generative A.I. can be an incredibly powerful tool, it also poses an existential risk to any business that doesn't handle it with a high level of care. If you're already using generative A.I. in your business model, or are thinking about implementing the tech, these are some steps you can take to safeguard your business without giving up the benefits of generative A.I.