The White House chief science officer points to cooperation between the United States and China on AI safety

Stay informed with free updates

Simply sign up artificial intelligence myFT Digest – delivered straight to your inbox.

The White House's chief science adviser said the United States will work with China on the safety of artificial intelligence systems in the coming months, signaling rare cooperation between the two powers.

Arati Prabhakar, director of the White House Office of Science and Technology Policy, told the Financial Times that despite trade tensions between China and the United States over artificial intelligence, countries will work together to reduce its risks and assess its capabilities.

“Steps have been taken to participate in this process,” Prabhakar said of future cooperation with China in the field of artificial intelligence. “We have to try to work [with Beijing]”.

Her comments are a clear indication that the two powers plan to cooperate in protecting the rapidly developing technology, even at a time when trade tensions between the two countries are rising.

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

“We are in a moment where everyone realizes that AI is the most powerful technology… AI is a very smart tool,” said Prabhakar, who is advising Biden on issues including regulating AI. “Every country is preparing to use it to build a future that reflects their values.”

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

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

Rapid progress in artificial intelligence has led to global concerns that it will exacerbate vulnerabilities to cyber attacks and the spread of misinformation.

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

“We are beginning to realize globally that the tools for evaluating AI models — to understand how effective they are, how safe and trustworthy they are — are very weak today,” she said.

The comments come as the United States has imposed export controls on chips and equipment to hinder China's ability to develop advanced technologies such as artificial intelligence.

Leading American and Chinese technology companies are also racing to build products powered by generative AI — models that can disseminate human-like text and materials in seconds.

Countries have chosen to organize their domestic AI clusters in different ways. China, for example, has provided detailed guidance on the development of artificial intelligence, emphasizing the need to control content.

By contrast, in October last year Biden issued a sweeping executive order on artificial intelligence to address threats, with a focus on national security and consumer privacy.

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

She said that the United States does not intend to slow down the development of artificial intelligence, but rather to continue to supervise the technology. However, the White House has faced criticism for jeopardizing the United States' competitive advantage in AI by moving too quickly to regulate it, though no specific legislation has been passed on the issue.

“In the United States, we recognize that we are at a moment when American leadership in the world depends on American leadership in artificial intelligence,” Prabhakar said.

However, she said that even US AI companies have accepted the need for clear methods for understanding and evaluating AI, which will be crucial to consumer confidence and business adoption. “Technology drivers within companies are among the loudest advocates of regulation,” Prabhakar said.

Ryan Haass, head of the China Center at the Brookings Institution, said that AI cooperation “is not good faith or benevolence; “It is a difficult calculation.”

He added: “Given how quickly technology is developing, there is an incentive to try to do something now. It will be important for both sides to demonstrate early that dialogue can lead to results.

Additional reporting by Dmitry Sevastopoulou in Washington