EU and US edge closer together on AI at latest Trade and Technology Council meeting

The EU and the US have edged closer to a common approach to artificial intelligence (AI) following the latest meeting of the transatlantic Trade and Technology Council (TTC), although their domestic strategies for regulating the technology remain poles apart.  

After meeting in the Parisian suburb of Saclay on 15-16 May, officials from Washington and Brussels also pledged to make it easier for academics to delve into social media platform data, and said they would better coordinate R&D efforts to wean themselves off Chinese supply chains.

One of the key outcomes was a commitment to a “joint roadmap on evaluation and measurement tools for trustworthy AI and risk management,” a step beyond what was agreed last September at the council’s inaugural meeting in Pittsburgh.

This latest meeting marks an intensification of cooperation on AI between the US and EU, said Sebastien Krier, an AI policy expert at Stanford University. “It goes a bit further than broad high-level commitments,” he said.

Measurement tools for “trustworthy AI” mean creating ways of checking things like error rates or ways of testing how fair an AI system is, said Krier. As for AI “risk management”, this means working out exactly how creators of AI systems account for different forms of risk – be it discrimination against minorities, environmental harm, or longer-term risks.

The statement doesn’t say that the EU and US will necessarily use the same tools to scrutinise their AI systems, but it does confirm that Washington and Brussels will work towards “interoperable approaches for managing AI risks.” They also agreed on a “shared hub/repository of metrics and methodologies for measuring AI trustworthiness and AI risks.”

Still, the EU and US are taking very different approaches to AI domestically. The EU’s AI Act, currently being scrutinised by lawmakers, would largely prohibit certain uses of AI, like real-time facial recognition.

But the US, at least at a federal level, is for now taking a voluntary approach. It has asked the National Institute of Standards and Technology to draw up guidelines for how to manage AI risks, leaving judgement calls on risk in the hands of companies that create AI systems. The White House is also working up a Bill of Rights for an Automated Society, in consultation with the public.

“The EU has been faster in building legislation, while the US is being more cautious,” said Krier.

Some campaigners want much more coordination between the EU and US, and for the council to take a much stronger line against specific uses of AI.

“The TTC must move purposefully to address the far-reaching problems of AI bias in the digital economy,” said Marc Rotenberg, president of the Center for AI and Digital Policy, a Washington DC-based think tank.

“The TTC should also draw red lines for those AI applications, such as facial surveillance, that should simply be prohibited. And more could be done to promote a dialogue between Congress and the [European] Parliament,” he said.

Social media access

The TTC also announced that it wants social media companies to grant academics more access to their data so that researchers can better study things like misinformation and online hate.

“We intend to advance modalities for data access for researchers to enable online platforms to grant such access with a view to advancing the common understanding of the societal risks applicable to specific online platforms,” the conclusions say.

This could start to address a long-running grievance of social scientists, who say certain social media platforms, particularly Facebook and Instagram, have in recent years choked off independent access, making scholarship increasingly difficult.

Facebook cut off access to a key way for researchers to monitor pages in 2019, noted Richard Rogers, a professor of new media and digital culture at the University of Amsterdam.

Facebook justified the move at the time by saying it was a response to privacy problems, including the Cambridge Analytica scandal.

Since then, researchers have tried workarounds using third party provided information, or Facebook’s own scheme to give academics access, called Social Science One.

But academics are still limited in what they can research. “We don’t know what’s being deleted or cleaned up,” said Rogers. That makes it very difficult to understand Facebook moderation policies. “It’s doing social media data research on their terms.” But other social media platforms, in particular Twitter, Reddit, and Telegram channels, are much more open to data gathering, Rogers said.

More R&D coordination

In the conclusions of the TTC, the EU and US also agreed to better coordinate their R&D strategies in several areas where they see China as a threat to their supply chains and technological dominance.

Washington and Brussels have long been worried about a near Chinese monopoly over the production of rare earth magnets, essential components in hard disks, smartphones, wind turbines, electric vehicles and more. The importance of these magnets is only going to increase as renewable energy technologies drive up demand.

To combat this risk, the EU and US will cooperate on establishing common priorities in the field of research and development, the conclusions say.

The same goes for communication technology. On 5G and 6G wireless, Washington and Brussels will work “towards a common vision and roadmap outlining some of the key challenges and needs of future generations of communication technologies,” the document says.

On solar panels, another area where the EU and US worry about dependence on China, there is also a promise to coordinate policy measures and incentives.

In addition, the TTC conclusions announced new research aimed at combatting authoritarian control over the internet. The EU and US will pursue joint research on the impact of internet shutdowns, the council agreed, and bring together social and political science researchers from both sides of the Atlantic to discuss the causes and impact of the manipulation of information by foreign governments.