1. The big hardware push
The government announced a £1.1bn investment into AI hardware – the cutting-edge semiconductor chips on which AI models such as ChatCPT and Claude run. The ambition of this spend is significant: “Build globally competitive AI hardware companies in the UK.”
The reality will be more complex. Right now, almost all advanced AI chips are made by one producer: Taiwan Semiconductor Manufacturing Corporation (TSMC). US companies such as Nvidia or Google do not produce AI chips themselves – they send hi-tech chip designs to TSMC, which carves these into silicon wafers using, among other things, deep ultraviolet lasers. It costs TSMC tens of billions to build a single chip foundry, and £1.1bn will not suffice to construct one on UK shores.
What the money could do is bolster domestic chip designers. Arm Holdings, based in Cambridge but listed in New York, is one of these – the government vaguely says it will have a “strategic industry partnership” with the company. The strongest component of the spending package appears to be a £400m procurement opportunity for UK chip makers – an encouraging sum, but industry experts say that a large part of this money was already announced in previous years. The announcement is carefully worded; the money will create opportunity, “including” for UK firms.
“It’s genuinely encouraging to see government treating AI compute as national infrastructure and putting real procurement weight behind it,” said Mark Boost, the chief executive of Civo, a UK-based cloud computing platform.
“My concern is that beneath the sovereignty language, the default flow of this money is likely to go to the usual suspects. Unless the contracts are structured deliberately, we’ll have spent a billion pounds building British-branded infrastructure on somebody else’s silicon, integrated by the established overseas vendors and rented from hyperscalers.”
2. AI skills and company adoption
The government made announcements on up-skilling – helping people learn how to use AI in their jobs – and getting companies to integrate AI systems into how they operate. This included committing £20m to an effort to map how AI is changing entry-level work, and developing practical advice for businesses to redesign roles.
A “bridge AI” scheme will give British companies funds to buy Uk-developed AI products, while there will also be an expansion of the UK’s “tech town” programme which has been pioneered by Barnsley. The government published bespoke plans for important sectors such as advanced manufacturing and the creative industries to adopt AI.
Bouke Klein Teeselink, an academic at King’s College London, says: “Very few people I know are using these tools to their full potential. So there is clearly a lot of productivity that has been left on the table in the UK.”
He adds, though, that ultimately it will be the private sector that embraces AI most efficiently rather than any government-backed programme that might move too slowly.
3. AI defence and US chip investments
Britain’s chief of defence staff, Sir Richard Knighton, announced the Rapid AI Delivery Taskforce, or RAID, which will help develop new AI models for the UK’s defence ecosystem. But, he says “the UK’s policy remains that humans, not machines, are accountable for decisions”.
Two big tech companies, AMD and Nebius, announced investments into the UK. AMD said it is putting “up to £2bn” to “accelerate AI innovation and research” in partnership with the University of Cambridge and Imperial College, among others. Nebius has said it will commit “approximately £1.7bn” to build out AI infrastructure in sites across the UK – this investment appears to actually be Nvidia chips.
4. Ordering Apple and Google to tackle nudity
The government said that big tech providers, including Apple and Google, must find ways to “detect and block nude images for children” via technical solutions or built-in features on tablets and smartphones. If they do not, they will be subject to criminal liabilities and fines.
This is a big one – and a far thornier demand than other requests the government made earlier this year, for example, requiring large tech companies to remove non-consensual nudes.
Right now, companies such as Google and Meta stop the spread of terrorist content, child sexual abuse material and non-consensual nudes by using a sort of digital watermarking system that flags this content. This system involves comparing newly posted content against an existing database of banned material, using these watermarks. It relies on users reporting problematic content, which is added to the database.
To detect and block nude images for children will probably require tech companies to take far more invasive measures. They will have to verify the ages of all their users – and likely, therefore, their identities as well. Second, they may have to scan far more extensively the content those users share between themselves. It could be very difficult for companies such as Google to do this without storing large amounts of information on their users. This information could be leaked, or under a different government, subpoenaed.
Companies that care about privacy – such as Signal, the encrypted messenger and Mullvad, a VPN provider – have sounded the alarm about this measure. Signal has said it will usher in a “dystopian combination of age verification and content scanning”, saying it could lead to “mass censorship capabilities”. It and Mullvad point out that fundamental privacies on the internet are what make it possible for activist groups to organise against authoritarian regimes.
Along with preventing children from sharing nudes, the government is expected to announce an under-16 ban on high-risk social media apps next week. It is not known which apps will be covered, but a blanket ban in Australia covers Facebook, Instagram, X, TikTok and Snapchat, among others. Features used by “safe” apps, such as disappearing messages, also face age restrictions.
An age ban will also raise the question of age verification – how will big tech companies know their users are children? At issue are questions such as: should big tech companies ask for – and potentially retain information on – users’ government-issued IDs? If so, what should be done about security, and about child users who do not have IDs yet?
Companies such as Google and Meta are working on this problem; Meta has said that age verification would be most effective if it is conducted by app stores, run by Google and Apple. Google, meanwhile, is working on a set of measures on “age assurance” – this could involve default guardrails for all users it does not know are adult, and at times asking users for their IDs. Other ways to solve the problem might also be possible: for example AI-powered facial age estimation, although this has reportedly been circumvented by children in Britain painting on moustaches.
