Posted by 
| SNEAK PEEK |
— Where does the Paris summit leave the U.K.’s AI policy?
— MPs tackle the data bill, amendments and all.
— Patrick Vallance sets out his plan to remake science.
Good morning and welcome to Wednesday,
This is Tom, heading back to London from a gloomy City of Light.
Come and say hi if you’re at the AI Fringe today.You can get in touch with your news, tips and views by emailing Tom Bristow and Laurie Clarke. You can also follow us on Twitter @TomSBristow and @llaurieclarke.
| DRIVING THE DAY |
THESE ARE THE TERMS: U.K. policymakers are combing over JD Vance’s speech at the AI Action Summit, in which he set out the Trump administration’s terms to allies: We’ll partner with you, if you deregulate. But there’s plenty else to unpick from the last two days — not least Britain’s reasoning for snubbing the summit’s final statement.
The official line: A No.10 spokesperson denied it was about cozying up to the U.S. or being rude to the French, but the U.K.’s “national interest.” The declaration “did not reflect the U.K.’s policy position on opportunity or security” and “didn’t provide enough practical clarity on global governance, nor sufficiently address harder questions around national security and the challenge AI poses to it,” they said.
Head scratching: But Michael Birtwistle, associate director at the Ada Lovelace Institute, said it was “difficult to pinpoint what exactly in that statement the government disagrees with.” Linda Griffin, VP for global policy at Mozilla, said: “It is unclear what the U.K. is aiming for.”
Two reasons: The Oxford Internet Institute’s Keegan McBride offered two possible explanations. First, the U.K. wants to align as closely as it can with the U.S. — not just for political reasons, but to ensure the AI Safety Institute continues to have American buy-in. There was likely also “some annoyance… at how France co-opted the safety series to turn it into a glorified trade show,” he said. The declaration barely acknowledged that this week’s summit came from safety efforts organized by the U.K. at Bletchley and Seoul.
Cheeky: “We’ve reduced references to the Bletchley Summit to the minimum,” a French official admitted, in an indication of how political concerns have pivoted from safety to winning the AI race. That change went down well with industry. OpenAI’s Sam Altman noted there was more talk of making AI faster and cheaper than making it safer in Paris.
On the other hand: But Anthropic’s Dario Amodei bucked the trend, declaring it “a missed opportunity” to address AI’s security and labor impacts and keep democracies in the lead. Yoshua Bengio agreed.
With us or against: Vance had opened his speech by saying: “I’m not here this morning to talk about AI safety,” later adding: “The Trump administration will ensure that the most powerful AI systems are built in the U.S.” but does not “want or need to go it alone.”
Mixed messages: However, keeping in lockstep with the U.S. appears hard to square with comments Kyle gave to the Times that the government’s AI bill will force American tech firms to give the U.K. government access for testing.
Are you sure? Those comments, which seemingly contradicted what Kyle had told journalists hours earlier, mark a return to language not heard since Trump’s election. Speaking to MPs yesterday, Science Minister Patrick Vallance glossed over a question on testing, adding: “There is no bill,” but only “a series of discussions around what this might look like.”
American werewolf in Paris: Kyle’s comments also contradict what the U.S. VP said he wanted from allies. “The Trump administration is troubled by reports that some foreign governments are considering tightening the screws on U.S. tech companies with international footprints,” Vance said. “America cannot and will not accept that, and we think it’s a terrible mistake.”
Ditch those regs: Vance proceeded to slag off GDPR and the EU’s digital rulebooks (the Online Safety Act was spared, and U.K. ministers say it won’t be reopened) and warned allies of working with China (he reportedly walked out of Monday’s dinner over comments by China’s vice-premier). But he was more nuanced on competition, pledging to give “Big Tech [and] little tech… a level playing field.”
The good: Vance’s invocation not to “strangle” AI concluded the main action of the summit, which benefitted from an endless supply of excellent food and coffee and a spectacular (if cold) venue at the Grand Palais.
The bad: “This is a trade fair,” messaged one industry rep on arrival, and it definitely had those vibes. Plus, the organization was poor, the venue was so vast it often appeared empty, and there were restrictions on getting into panels. Good luck to India for the next one!
| AGENDA |
MINISTERS GRILLED: In a big week for tech policy, MPs get the chance to quiz DSIT ministers in the Commons at 11:30 a.m. this morning. Tabled questions cover AI Growth Zones, public services, online pornography, quantum and, of course, rural broadband — but don’t be surprised if hot topics like DeepSeek and the U.K.’s Paris snub come up at topicals.
DATA DEBUT: After PMQs, the Data (Use and Access) Bill gets its first airing in the Commons. There are no amendments, but MPs could tackle tricky issues like automated decision-making and — thanks to a government defeat in the Lords — AI and copyright.
Prior reading: MPs may want to prepare by reading this briefing on how the bill has changed in the Lords. On Monday, the Information Commissioner published his own opinion on those changes, confirming: “overall, the Bill remains one which I support,” but raising concerns on that pesky copyright amendment, as well as government additions on children’s data and deepfakes. The right balance on automated decision-making is “ultimately… a decision for Parliament,” he noted.
AT THE FRINGE: The AI summit is over, but the AI Fringe continues. Today’s panels cover AI’s supply chains, sustainability, “slop” and deepfakes. It closes with a session reflecting on the Paris summit with MTUK’s Tom Bristow.
PARTY TIME: AI Minister Feryal Clark is expected to drop in at the opening of London’s answer to Station F, the London AI Hub, this evening, along with the project’s backers.
| AROUND THE WORLD |
HIGH STAKES: European Commission President Ursula von der Leyen unveiled a €20 billion fund to build “AI GigaFactories” and support a “European brand of AI” based on “openness”.
IN COMPARISON: Saudi Arabia announced $15 billion for AI at its own conference in Riyadh.
CLOSER LOOK: European privacy watchdogs are expanding a probe into ChatGPT to include DeepSeek, Reuters reports.
BRIDGING DIVIDES: Apple will partner with Alibaba for AI services in China, SCMP reports.
COURT DRAMA: A U.S. federal judge rejected a “fair use” defense for AI training in a copyright case brought by Thomson Reuters.
IMPACT: U.S. aid cuts are impacting efforts to tackle online child abuse, Wired reports.
| ARTIFICIAL INTELLIGENCE |
DECLARATION SNUB: “With these summits, there are always conversations and discussions going right to the 11th hour,” AI Minister Feryal Clark told the AI Fringe yesterday, explaining why the U.K. failed to join 60 other countries and international bodies in signing the Paris summit’s final statement.
Tried, promise: “We work really closely with France … they’re an important ally when it comes to AI, and we did sign up to a lot of the other things that were set out at the summit,” Clark said. (The U.K. joined a “Coalition for Sustainable AI” and signed a statement on cybersecurity yesterday.) But “there were areas within the agreement that we didn’t quite agree with, so that’s why we didn’t sign.”
Not a diary issue That’s despite several meetings between Clark and her French counterparts in recent weeks. “I was especially interested to hear about my French counterparts’ proposals around how they were looking to diffuse AI across their public services and their plans to to utilize the opportunities that AI has for transforming public services.”
Top priorities: On her own priorities, Clark name-checked AI Growth Zones and “a responsible AI development board,” bringing together civil society, unions and experts. “We spent the whole past couple of years talking about how terrifying AI was … and now we’ve got a new government who are like … isn’t it all amazing … How do we ensure that our citizens have a voice and say in shaping what is happening?”
| RESEARCH |
VALLANCE’S VISION: Patrick Vallance’s appearance before the Commons science and tech committee yesterday provided a window into his vision to embed science and tech across government — and several chances for George Freeman to remind us that he once held the role himself.
First, numbers: Vallance said DSIT’s £13.9 billion R&D budget for 2025/26 was a “good” result “hard won from the chancellor.” “What we need to ensure is that we have the best possible use of that money to get the maximum effect,” he said.
Next, balance: That comes from a balance of “basic, curiosity-driven research”, “applied research” aligned to the government’s agenda, and supporting businesses to scale and invest. “You can’t be an economy that grows faster than average if you don’t have science and technology companies at the very heart of it,” Vallance said.
Tough gig: Finding that balance will fall to new CEOs at URKI and Innovate UK, who will be named “shortly,” Vallance said. They’ll be asked to be clearer about how their work falls into the three categories and deliver returns, while making URKI “lighter touch, more strategic.” The “exciting” ARIA is already delivering results and should remain independent.
More job news: Also to be announced “very shortly” is a chair for the Regulatory Innovation Office, charged with joining-up regulators and boosting capacity. It’s starting with four priority areas — but Vallance sounded confident it’ll expand.
Pet projects: Another Vallance brainchild is a £25 million R&D Missions fund announced at the budget. Government mission boards will be able to tap into it to commission research into their challenges, and Vallance said it aims to make “using science and technology a routine in government.”
What else? That adds to a to-do-list that Vallance said also includes procurement reform, working out how to “extract value” from public data, scoping public infrastructure projects and deciding which research areas should get 10-year budgets. More broadly, Vallance said he wants to ensure science and tech is “informing all aspects” of the industrial strategy. DSIT itself is leading on two sectoral plans, as well as a cross-cutting piece on innovation.
How’s the new role? Asked about his new gig championing the Oxford-Cambridge corridor, Vallance said it would be a key test of “joined-up government.” He also insisted that success wouldn’t have to come at the expense of other parts of the country. But “as a nation, we’ll be doing ourselves a massive disservice if we don’t deliver on this.”
Looking abroad: Vallance said he was “very pleased” to hear the chancellor’s recent support for ensuring the visa system welcomes top scientists and engineers. The U.K.’s re-integration into Horizon has also been a success, Vallance said, and downplayed the impact of R&D funding cuts in the U.S. But there are “challenges” elsewhere, he admitted, particularly China.
On AI: Vallance said the AI Safety Institute had considerable expertise and would have a “big role” in future AI legislation — but currently “there is no bill” to speak of. When it does arrive following consultation, it will be focused on the frontier, he said. The U.K. is taking a “different” approach to the EU, Vallance said, but predicted convergence over time. “It’s pretty pointless ending up with completely bespoke regulation in one country, because this is a global area.”
| BEFORE YOU GO |
HOMEWORK, PLEASE: TechUK published its suggestions for the spending review, expected in June. The industry body wants the chancellor to raise the R&D target to 3 percent of GDP by 2029, a funded and detailed roadmap to deliver on the AI Opportunities Action Plan, and ring-fenced cash for SME digitization, among much else.
Final warning: It also gave the chancellor a ticking off for her “negative” autumn budget tax rises. It urges government to work in “partnership” with the tech sector, and warns that “failing to seize the moment risks resurrecting familiar challenges” of underinvestment.
MR ROBOTO: DSIT will consider a new roadmap to “position the U.K. as a global leader in robotics and Smart Machines” by 2035 as part of the industrial strategy and cross-government review of technology adoption.
CASE STUDY: The Digital Regulation Cooperation Forum shared the first case study from its AI and Digital Hub advisory service, with more to come. The pilot closes on Feb. 28.
WHAT WE’RE DOING: On Safer Internet Day, Online Safety Minister Maggie Jones wrote about what the government’s doing to keep children safe for Netmums.
NOT A GOOD LOOK: TikTok may have deleted data related to four deceased British children whose families are suing TikTok, an executive told the BBC, blaming data rules.
SNOOPER SNAFU: The Home Office failed to maintain a secretive board that could play a pivotal role in deciding whether Apple must weaken encryption, ComputerWeekly reports.
HALLUCINATION WARNING: Research by the BBC has found that AI chatbots generate “significant inaccuracies and distorted content from the BBC.”
Morning Technology UK wouldn’t happen without the production team.
