The Federal Trade Commission is investigating OpenAI over whether ChatGPT poses a threat to consumers by infringing on their privacy and promoting false information.
The investigation appears to be “expansive,” according to the The Washington Post, which cites a 20-page letter the US regulator sent to OpenAI this week.
The letter says the FTC is probing the San Francisco company because its large language models (LLMs), which power ChatGPT, may have engaged in “unfair or deceptive” privacy practices while posing “reputational harm” to consumers.
The document includes 49 questions—many of them involving confidential matters—including what data the company uses to train ChatGPT, how the data is obtained, and the company’s safeguards to prevent the chatbot from learning users’ private information.
The FTC is also asking OpenAI for more details about a software bug that caused ChatGPT’s website to briefly leak users’ conversation histories and the payment details of paid users. The letter notes OpenAI must suspend any procedures involving document destruction. In addition, the FTC’s investigation will consider whether “monetary relief would be in the public interest.”
The FTC declined to comment. “FTC investigations are nonpublic so we generally do not comment on whether we are investigating a particular matter,” a spokesperson told PCMag.
OpenAI didn’t immediately respond to a request for comment. But the investigation comes amid a flurry of lawsuits against chatbot developers over the same privacy and defamation concerns.
Although ChatGPT can provide accurate answers on a huge range of topics, it can also “hallucinate” and supply incorrect replies, potentially misleading users. The other concern is that OpenAI and other companies scrape the web, including collecting people’s social media posts without their consent, to train and refine their chatbots. Hence, the FTC’s investigation could expose OpenAI’s internal practices and put the company under greater regulatory scrutiny.
Enza Iannopollo, an analyst at research firm Forrester, noted the lack of rules around LLMs could prove to be a "a complete disaster" for AI chatbots due to the potential regulatory fines, investigations, and harms to users. "As long as large language models (LLMs) remain opaque and rely largely on scraped data for training, the risks of privacy abuses and harm to individuals will continue to grow," she said in an email.