The US Federal Trade Commission officially investigates OpenAI, and the regulatory storm is coming?

The U.S. Federal Trade Commission (FTC) has officially launched an investigation into OpenAI, the company behind the artificial intelligence chatbot ChatGPT.

According to the content of the letter exposed by the US media on the 13th local time, the FTC informed OpenAI that it will investigate whether ChatGPT harmed people's interests by publishing false information, and whether OpenAI has "unfair or deceptive" privacy and data security practices.

In its letter, the FTC also asked OpenAI to share internal information, including how the company retains user information and the steps the company has taken to address the risk of "false, misleading, or defamatory" claims from its models.

This is not without warning. In May, the White House met with CEOs of OpenAI, Google, Microsoft and others to discuss the risks of related technologies. The White House also said that the seven largest AI companies have agreed to open their models to a degree of public scrutiny at the annual Def Con Geek Conference in August.

Regarding the disclosure of models, Theos Evgeniou, a professor of decision-making science and technology management at INSAED, told First Financial Reporter: "I think the discussion on sharing model information is the wrong focus. Assessing whether a model is trustworthy should only be based on whether it is safe or not, not by examining the structure of the model itself, which can be complex. This can be achieved by studying the behavior of the model. I think a model can solve a lot of problems without much disclosure. ”

What the FTC wants to regulate

On the 13th, FTC Chairman Lina Khan declined to comment on the investigation at a hearing of the House Judiciary Committee, but she said that regulators' concerns include that ChatGPT and other AI services are "being fed a lot of data" and that these AI companies "did not check what type of data was entered."

She added: "We've heard cases where some sensitive information was leaked when (AI chatbots) responded to inquiries. We also hear slander, defamatory statements, things that are completely untrue. It is this fraud and deception that we are worried about. ”

Some users reported that ChatGPT fabricated names, dates and facts, as well as forged links to news websites and references to academic papers. This problem is known in the industry as "hallucinations."

Regulators have also expressed concern about the vast amount of data grabbed by the language model behind ChatGPT. Two months after OpenAI's launch, it has more than 100 million monthly active users. Microsoft's new Bing search engine, which also uses OpenAI, was used by more than 1 million people in 169 countries within two weeks of its launch in January.

In its civil subpoena to OpenAI, the FTC asked the company in detail about its data security measures, marketing efforts, practices for training AI models, and handling of users' personal information. For example, what is the supervision of human reviewers who directly affect consumers, as well as information on consumers' past complaints.

OpenAI CEO Sam Altman said on Twitter late on the 13th that the company will cooperate with the FTC and believe that the company's technology is safe and beneficial to consumers. "We protect user privacy, and we design this system to understand the world, not individuals." But he acknowledges: "We're very honest about our technical limitations, especially when we have shortcomings." Our limited profit structure means we have no incentive to earn unlimited returns. ”

According to ChatGPT, "limited profit" means that OpenAI puts a social and ethical mission ahead of the pursuit of financial goals.

Future regulatory direction in the United States

Since the beginning of this year, with the rapid rise of generative AI applications, related AI products have become the focus of global regulators.

The Biden administration has begun looking at the need for a review of AI tools such as ChatGPT. The Commerce Department formally solicited comments in April on so-called accountability measures. The White House Office of Science and Technology Policy is also developing strategies to clarify the benefits and drawbacks of AI, including increased hacking capabilities, discriminatory decision-making by AI systems, and the potential for AI-generated content to undermine elections.

In May, the FTC issued a warning to the AI industry, saying it is closely monitoring companies' choice and use of AI technology and the substantial impact on consumers.

Under Senate Majority Leader Chuck Schumer, lawmakers from both parties have also made AI regulation a priority in this Congress. But new legislation or measures may take at least several months to be introduced. Lawmakers are also concerned that the legislation could slow the pace of innovation in the United States.

Regarding the trend of regulating AI in the United States, Efgenio explained to First Finance and Economics: "Business needs supervision. Even tech executives are calling for regulation. This is how Altman stood at the U.S. Senate hearing in May. Because they know that without regulation, people will distrust technology. ”

But Evgenio doesn't think U.S. regulation will be too hasty or harsh. "The way America likes it is: don't rush, see how it develops, allow innovation first, and solve it when there is a problem." This is because the United States is also often very quick when deciding to act. We've seen how amazing the speed with which things have changed in the United States after the financial crisis has been. This is also in line with the willingness of the United States to let go and observe. He said.

FTC "nosy"?

The FTC is the corporate regulator responsible for consumer protection in the U.S. federal government, and can regulate all unfair and deceptive business practices that may harm consumers, as well as unfair competition practices.

But critics argue that under Khan, the agency sometimes oversized its remit.

Adam Kovacevich, founder of the Chamber of Progress, a lobbying group representing the tech industry, said: "When ChatGPT says the wrong thing about someone and could damage their reputation, does that fall under the jurisdiction of the FTC? I don't think it's clear at all. Such issues, he argued, were "more of a rhetoric than their purview."

In addition, at this Thursday's hearing, Khan was accused of "harassing" the Twitter company. Twitter said in a court filing that it was subject to a "cumbersome and vexatious enforcement investigation" by the FTC. Khan responded by saying the agency was only interested in protecting user privacy and that "we are doing everything we can to ensure Twitter complies with orders."

The FTC's troubles don't stop there. This week, the FTC's lawsuit to block Microsoft's acquisition of Activision Blizzard was also dismissed by a U.S. federal court. On the 13th, the FTC appealed the ruling.