The Federal Trade Commission is investigating OpenAI's flagship chatbot in a wide-ranging probe that is looking into potential unfair and deceptive practices around privacy, data security and consumer harm, according to various reports.
In a letter obtained by the Washington Post, the FTC requested details about OpenAI's large language model products, such as ChatGPT, including the company's marketing practices, third-party access, data collection and how it assesses risk and safety.
The letter also requests details about how the San Francisco company handles potentially "false, misleading, disparaging, or harmful statements" that its products might generate about real people, and the steps it takes to "address or mitigate" that risk.
And it cites a security incident that OpenAI previously disclosed involving a bug that potentially exposed private user chat histories and payment information. The letter also requested information about any other security incidents that OpenAI was aware of.
In March, OpenAI published a blog post acknowledging that it took ChatGPT offline "due to a bug in an open-source library" that exposed some users' chat data and may have exposed some payment information from subscribers.
The FTC declined to comment, and OpenAI CEO Sam Altman took to Twitter Thursday afternoon to express his disappointment at the investigation being leaked.
"It is very disappointing to see the FTC's request start with a leak and does not help build trust. that said, 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," Altman tweeted.
FTC Chair Lina Khan was coincidentally testifying at a House Judiciary Committee hearing about unrelated matters, but lawmakers also questioned her about the agency's investigation into OpenAI, the Washington Post reported.
“We’re focused on, ‘Is there substantial injury to people?’ Injury can look like all sorts of things," Khan said in response to a line of questioning from North Carolina Republican Rep. Dan Bishop about whether the FTC has the authority to launch this investigation.
OpenAI publicly launched ChatGPT in November, catapulting generative AI into the mainstream and sparking a global arms race to develop and invest in the technology.
It also teased its image generator, which it calls DALL-E 2, months earlier and made it publicly available in September.
Italian regulators temporarily blocked ChatGPT after the company disclosed its data breach in March, the Associated Press reported at the time, but reversed the decision about a month later after OpenAI complied with its requests.
On Monday, comedian Sarah Silverman and other plaintiffs sued OpenAI and Meta for using the copyrighted content without permission to train their large language models. And in a separate lawsuit filed earlier this month, two authors also alleged that OpenAI illegally used their copyrighted materials, the Guardian reported.