Rishi Sunak announces agreement with tech firms to vet AI models before release

Rishi Sunak has said Britain’s artificial intelligence (AI) safety summit will “tip the balance in favour of humanity” after reaching an agreement with technology firms to vet their models before their release.

The Prime Minister said “binding requirements” are likely to be needed to regulate the technology, but now is the time to move quickly without laws.

He described the UK’s summit held at Bletchley Park, the home of Allied codebreaking during the Second World War, as “only the beginning of the conversation”.

The Prime Minister has previously said he would not want to “rush to regulate” the technology with binding rules, but in a press conference at the close of the event on Thursday he indicated it would likely need to be put on a statutory footing in the future.

A statement agreed by governments and tech companies at the summit acknowledges that both parties should have a role in ensuring external safety testing of powerful “frontier” AI models is carried out, rather than firms having sole responsibility.

Asked whether the non-binding agreement reached with tech firms was enough to mitigate risks without the backing of legislation, Mr Sunak replied: “The lesson is that we need to move quickly and that’s what we’re doing.

“The technology is developing at such a pace that governments have to make sure that we can keep up.

“Now, before you start mandating things and legislating for things, I think… that takes time and we need to move faster, and we are, but secondly, you need to know exactly what you’re legislating for and that’s why our safety institute is so important.

“So far we’ve got the co-operation we need, but, of course, I think everyone would acknowledge, ultimately, binding requirements will likely be necessary, but it’s important that we do those in the right way and that needs to be based on empirical evidence that we’ll get from our testing.”

He also had warm words for Elon Musk, saying he was “delighted” that “one of the leading actors on AI” had attended the summit – although the tech billionaire posted a picture appearing to criticise the gathering just as Mr Sunak started delivering his closing remarks.

The X owner, with whom the Prime Minister will hold a televised talk on the technology later, shared a cartoon image on his social media platform of animals appearing to depict representatives from the UK, EU, US and China at the summit, saying: “We declare that AI poses potentially a catastrophic risk to humankind and I cannot wait to develop it first.”

Mr Musk earlier described AI as “one of the biggest threats” facing humanity and it is “not clear to (him) if we can control such a thing” when humans face “something that is going to be far more intelligent than us”.

Asked whether the conversation between the two would not be livestreamed because Mr Sunak is worried about what the tech entrepreneur might say, the Prime Minister replied: “Elon Musk is someone who has for a long time spoken about AI… I’m delighted that he was attending and participating yesterday.”

But he also said the achievement of the summit had been to bring together governments and leaders across the world to discuss mitigating the risks of the technology rather than “focus(ing) on one individual”.

It came after the Technology Secretary said a Terminator-style rise of the machines was a “potential area” where AI development could lead – remarks Mr Sunak did not distance himself from when asked about them later.

Michelle Donelan was speaking to Times Radio from the summit in Milton Keynes, where ministers have convened governments from around the world alongside tech firms and civil society to discuss the risks of the advancing technology.

Asked about the possibility of a “Terminator scenario” – a reference to the Arnold Schwarzenegger film in which machines take over the world – she said: “Well, that is one potential area where it could lead but there are several stages before that.”

As he arrived at the summit on Thursday, Mr Sunak said “we can’t be certain” about the risks of AI but there is a possibility they could be on a similar scale to pandemics and nuclear war.

The Prime Minister said: “People developing this technology themselves have raised the risk that AI may pose and it’s important to not be alarmist about this.

“There’s debate about this topic. People in the industry themselves don’t agree and we can’t be certain.

“But there is a case to believe that it may pose a risk on a scale like pandemics and nuclear war, and that’s why, as leaders, we have a responsibility to act to take the steps to protect people, and that’s exactly what we’re doing.”

The summit has seen delegations from around the world, including the US and China, agree on the so-called “Bletchley declaration” – a statement on the risks surrounding the technology to be used as the starting point for a global conversation on its development.

Mr Sunak held a flurry a bilateral meetings with United Nations secretary-general Antonio Guterres, European Commission president Ursula von der Leyen and Italian Prime Minister Giorgia Meloni after arriving in Bletchley on Thursday.

Ms Meloni said she was “proud of (her) friendship” with her UK counterpart and hoped they could work together on AI to “solve the biggest challenge that maybe we have in this millennium”.

After the meetings, he sat down for a roundtable discussion with representatives from France, US, Germany, Australia, the EU and the UN including Kamala Harris, telling the US vice-president that her country’s executive order on AI, signed days before the summit, was “very welcome in this climate”.

The Prime Minister said: “I wanted us to have a session to talk about this issue as leaders with shared values in private and hear from all of you about what you’re most excited about, what you’re concerned about and how we can look back in five years time on this moment and know that we made the right choices to harness all the benefits of AI in a way that will be safe for our communities but deliver enormous potential as well.”

He later held a similar panel discussion with tech bosses including Sir Nick Clegg, president of global affairs at Meta, and representatives from Google, DeepMind and OpenAI.