Press play to listen to this article
Voiced by artificial intelligence.
LONDON — He’s the prime minister who cultivated a ‘tech bro’ image — even if the tech world is unconvinced by his credentials.
Now Rishi Sunak hopes to pitch the U.K. as a leader in global AI governance, leveraging the setting of standards as a soft power tool while boosting Britain’s emerging AI sector.
He may have a battle on his hands. For as Sunak flies to Washington this week for his first White House summit as U.K. prime minister — with the future of AI high on his agenda — he finds Britain already playing catch-up on the world stage.
Having left the European Union in January 2020, the U.K. is locked out of key forums between the EU and U.S. such as the Tech and Trade Council (TTC), where AI governance plans are negotiated on a bilateral basis. Britain’s requests for a similar dialogue with Washington have been repeatedly rebuffed, leaving Sunak forced to pursue direct channels like his White House summit with U.S. President Joe Biden this Thursday.
At the center of Sunak’s grand plan is an international AI summit this fall that would convene like-minded allies in London to discuss the risks of the technology and how best to regulate it, two Whitehall officials and one government adviser told POLITICO. The idea was floated at the prime minister’s Downing Street roundtable with the heads of OpenAI, Google DeepMind and Anthropic in late May.
Sunak has also expressed interest in the U.K. hosting a global watchdog on AI akin to the International Atomic Energy Agency (IAEA), an intergovernmental body which promotes peaceful uses of nuclear energy. An IAEA equivalent for AI is an idea that OpenAI’s Sam Altman has previously championed.
A “CERN for AI” is another of Sunak’s proposals, mooting an international research body akin to the one that exists for international particle physics. Sunak plans to air all three ideas in his talks with Biden this week.
Whether Biden is ready to listen remains to be seen.
Following last week’s fourth meeting of the TTC, the EU, U.S. and Canada have already agreed to develop an AI code of conduct between themselves. The voluntary agreement will be presented to other G7 countries — including Britain — at a fall conference planned in Japan, which holds the G7 presidency this year.
The U.K.’s AI offensive is therefore about trying to carve out a tech leadership role outside the TTC, the U.K. government adviser quoted above said.
No. 10 wants to create an influential forum that includes other unrepresented countries with advanced AI industries like Japan and South Korea, as well as the U.S. and EU27 nations.
Britain believes the EU and the U.S. are already creating systems and that “our role would be a broker or an amplifier, and also bringing in other parties, almost as a voice of the third countries outside those two blocs,” the adviser added. The U.K. government has expressed a desire to tread between the more hard-line EU approach, whose AI Act is set to come into force in the next few years, and the relatively hands-off attitude of the U.S.
The adviser said that Japan and Canada have been supportive of the U.K.’s summit ambitions, but that plans have not yet progressed far.
This big U.K. push on AI wasn’t always a given. One government adviser says Sunak selected artificial intelligence from a list that was presented to him by officials setting out a list of cutting-edge technologies where the U.K. could potentially play a leading role in international standards-setting. Also on the list — but ultimately rejected — were “semiconductors” and “space.”
It all represents a marked shift from April, when the U.K. government released an AI white paper amid much fanfare over its “pro-innovation” and light-touch approach.
Instead of setting up a dedicated regulator to deal with AI, as some had expected, the white paper advocates existing regulators fold AI into their remits where relevant.
But fast forward a few months and the global outlook on AI is a lot dimmer. Barely a week goes by without AI experts issuing catastrophic warnings about the technology’s potential impacts, and even AI companies themselves have expressed a desire for regulation.
The U.K.’s Department for Science, Innovation and Technology (DSIT) is now holding weekly meetings with ministers and the department’s permanent secretary, as well as other officials, to discuss the technology. Over in No. 10, Sunak’s closest aides, his chief of staff Liam Booth-Smith and deputy chief of staff Will Tanner, are now said to be heavily engaged with the issue.
Unlike No. 10, which has come late to the party on AI, other parts of the U.K.’s government machine — most notably the Alan Turing Institute, which is mostly funded by London and has garnered a global reputation for evidence-based research into thorny issues like AI ethics — have close ties to many of their counterparts around the globe.
As concern has grown, the U.K. government’s rhetoric has become much more cautious. Still, many experts have suggested that the AI white paper doesn’t adequately address the risks posed by foundation models like GPT-4 and that more regulation dealing directly with generative AI is probably required.
One industry representative told POLITICO that civil servants appeared to no longer be pushing the vision contained within the white paper, and that a sense of disorder was prevailing around the issue.
Despite this, government officials continue to insist the AI white paper won’t need to be dramatically updated. The government official quoted above told POLITICO that the only real shift in recent weeks was in emphasis — from ‘pro-innovation’ to ‘risk-assessment.’
Other industry representatives who have had dealings with government agree it’s unlikely that entirely new regulation will be created at this stage, although some expect more tools and funding to be handed to existing regulators.
Inside the operation
Set to take a leading role in settling the question of how the U.K. governs AI is whoever the government selects to head up its AI task force. Matt Clifford, also chair of ARIA, is currently interim chair and No. 10’s preferred candidate for the role, a second government official told POLITICO.
Clifford represents “the technologically-literate version of No. 10,” a third industry executive told POLITICO. At his side is University of Cambridge PhD student Nitarshan Rajkumar, who joined DSIT as a policy adviser on artificial intelligence in April.
The U.K.’s tech envoy, Joe White, is busy promoting the message to Silicon Valley companies that the U.K. wants to carve a path between the hard-line EU approach and hands-off U.S. attitude. The U.K.’s intention is to entice the U.S., as a powerful economic bloc, to sign up to its light-touch approach, in the hopes that this will help cement it as the international standard.
It remains to be seen if the U.K. can have a meaningful say in matters from its vantage point outside the EU. This week’s Washington trip will give Sunak a chance to prove that it can.
Mark Scott and Brendan Bordelon contributed reporting.