Biden Administration Weighs Possible Rules for AI Tools Like ChatGPT

The administration’s action comes amid a boom in the use of artificial-intelligence tools that can quickly generate humanlike writing, images, videos and more. ChatGPT, the chatbot from

Microsoft Corp.

MSFT -0.76%

-backed startup OpenAI, has been estimated by some analysts to have reached 100 million users faster than any consumer app in history. 

“It is amazing to see what these tools can do even in their relative infancy,” said

Alan Davidson,

who leads the National Telecommunications and Information Administration, the Commerce Department agency that put out the request for comment. “We know that we need to put some guardrails in place to make sure that they are being used responsibly.” 

The comments, which will be accepted over the next 60 days, will be used to help formulate advice to U.S. policy makers about how to approach AI, Mr. Davidson said. He added that his agency’s legal mandate involves advising the president on tech policy, rather than writing or enforcing regulations.  

Industry and government officials have expressed concern about a range of potential AI harms, including use of the technology to commit crimes or spread falsehoods. 

Children’s safety was top of mind for Sen. Michael Bennet (D., Colo.) when he wrote last month to several AI companies, asking about public experiments in which chatbots gave troubling advice to users posing as young people. 

“There are very active conversations ongoing about the explosive good and bad that AI could do,” said Sen. Richard Blumenthal (D., Conn.), in an interview. “This, for Congress, is the ultimate challenge—highly complex and technical, very significant stakes and tremendous urgency.” 

Tech leaders including

Elon Musk

recently called for a six-month moratorium on the development of systems more powerful than GPT-4, the version of OpenAI’s chatbot released about a month ago. They warned that an unfolding race between OpenAI and competitors such as

Alphabet Inc.’s

Google was occurring without adequate management and planning about the potential risks. 

President Biden discussed the topic with an advisory council of scientists at the White House last week. Asked by a reporter whether the technology is dangerous, Mr. Biden said: “It remains to be seen. It could be.”

Among those in the advisory group were representatives of Microsoft and Google. They and other companies releasing AI systems have said they are constantly updating safety guardrails, such as by programming chatbots not to answer certain questions. 

In some cases, companies have welcomed and sought to shape new regulations. 

“We believe that powerful AI systems should be subject to rigorous safety evaluations,” OpenAI said in a recent blog post. “Regulation is needed to ensure that such practices are adopted, and we actively engage with governments on the best form such regulation could take.”  

Rep. Nancy Mace (R., S.C.), chair of a House Oversight Committee panel on technology, last month opened a hearing on AI with a three-minute statement discussing AI’s risks and benefits. Then she added: “Everything I just said in my opening statement was, you guessed it, written by ChatGPT.” 

In her own words, Ms. Mace asked witnesses whether the industry is moving too fast with AI development. “Are we capable of developing AI that could pose a danger to humanity’s existence?” she said. “Or is that just science fiction?”

One of the witnesses,

Eric Schmidt,

the former Google chief executive who chaired a congressional commission on the national-security implications of AI, said “everyone believes that the AI we are building is what they see in the Terminator movies, and we are precisely not building those things.”

Mr. Schmidt said Congress should focus on fostering talent development and innovation. “Let American ingenuity, American scientists, the American government, American corporations invent this future, and we’ll get something pretty close to what we want,” he told lawmakers. “And then you guys can work on the edges, where you have misuse.” 

Rep.

Gerry Connolly

(D., Va.) expressed skepticism about Congress’s ability to adopt AI rules, pointing to its lack of action to address the “awesome power” that social-media companies wield. 

“Without any interference by the government, they make all kinds of massive decisions in terms of content, in terms of what will or won’t be allowed, in terms of who gets to use it,” he said. “Why should we believe that AI would be much different?”

Absent a federal law focused on AI systems, some government agencies have used other legal authorities. 

Financial-sector regulators have probed how lenders might be using AI to underwrite loans, with an eye toward preventing discrimination against minority groups. 

The Justice Department’s antitrust division has said it is monitoring competition in the sector, while the Federal Trade Commission has cautioned companies that they could face legal consequences for making false or unsubstantiated claims about AI products. 

In the public-comment document released Tuesday, the federal tech-advisory agency asked whether measures should be added to ensure public trust in AI systems, such as “quality assurance certifications.”

The document asked whether new laws or regulations should apply, but stopped short of detailing potential harms or endorsing any specific safeguards. 

Write to Ryan Tracy at [email protected]