White House science chief signals US-China co-operation on AI safety

Stay informed with free updates

Simply sign up to the Artificial intelligence myFT Digest — delivered directly to your inbox.

The US will work with China on the safety of artificial intelligence systems in the coming months, the White House’s top science adviser has said, signalling a rare co-operation between the two powers.

Arati Prabhakar, director of the White House Office of Science and Technology Policy, told the Financial Times that, despite Sino-US trade tensions over AI, the countries would work together to lessen its risks and assess its capabilities.

“Steps have been taken to engage in that process,” Prabhakar said of future collaboration with China on AI. “We have to try to work [with Beijing].”

Her comments are an explicit signal that the two powers plan to collaborate on safeguarding the rapidly developing technology, even at a time of heightened trade tensions between the countries.

In November, China signed the UK’s Bletchley Park agreement on standards for the technology, while US President Joe Biden and his Chinese counterpart Xi Jinping discussed working together on AI at a summit in California the same month.

“We are at a moment where everyone understands that AI is the most powerful technology . . . every country is bracing to use it to build a future that reflects their values,” said Prabhakar, who advises Biden on issues including AI regulation.

“But I think the one place we can all really agree is we want to have a technology base that is safe and effective,” she added. “So I think that is a good place for collaboration.”

Earlier this month, the Financial Times reported that leading US AI companies including OpenAI had engaged in covert meetings with Chinese experts to discuss the emerging risks associated with the technology.

Rapid advances in AI have led to global concern that it will exacerbate vulnerabilities to cyber attacks and spread misinformation.

Prabhakar said the best options available to evaluate the safety of new AI systems were “not up to scratch” because of the complex and opaque nature of the technology.

“We are starting to have a global understanding that the tools to assess AI models — to understand how effective, how safe and trustworthy they are — are very weak today,” she said.

The comments come as the US has imposed export controls on chips and equipment to hamper China’s ability to develop advanced technologies such as AI.

Leading US and Chinese tech companies are also racing to build products powered by generative AI — models that can spew out humanlike text and material in seconds.

The countries have chosen to regulate domestic AI groups in different ways. China, for instance, has provided detailed guidance on the development of AI, emphasising the need to control content.

By contrast, in October last year, Biden issued a sweeping executive order on AI to tackle threats, focused on national security and consumer privacy.

Prabhakar said that, while China and the US may disagree on certain values and approaches to regulation, “there will also be places where we can agree”, including on global technical and safety standards for AI software.

She said the US did not intend to slow down AI development, but to maintain oversight of the technology. However, the White House has faced criticism for imperilling US’s competitive advantage in AI by moving too swiftly to regulate it, even though no specific legislation has been passed on the issue.

“In the United States, we understand that we’re in a moment where American leadership in the world depends on American leadership in AI,” Prabhakar said.

However, she argued even US AI companies accepted the need for clear methods to understand and assess AI, which would be crucial to consumer confidence and business adoption. “The drivers of the technology within the companies are among the most vocal advocates for regulation,” said Prabhakar.

Ryan Hass, head of the China centre at the Brookings Institution, said co-operation on AI is “not goodwill or charity; it is a hard-nosed calculation”.

He added: “Given how fast the technology is developing, there is an impetus to try to do something now. It will be important for both sides early on to demonstrate that the dialogue can generate results.”

Additional reporting by Demetri Sevastopulo in Washington