The issue of AI sovereignty took center stage at London Tech Week, as the UK government unveiled a series of announcements aimed at asserting control over a technology dominated by the US and China. The event highlighted the government's growth narrative and assertive AI strategy, though some measures appeared to repackage existing commitments.
Major Hardware Investment
The government pledged £1.1 billion for AI hardware, specifically cutting-edge semiconductor chips used in models like ChatGPT and Claude. The stated ambition is to "build globally competitive AI hardware companies in the UK." However, the reality is complex. Almost all advanced AI chips are produced by Taiwan Semiconductor Manufacturing Corporation (TSMC). US firms like Nvidia and Google design chips but rely on TSMC for fabrication, which costs tens of billions per foundry. The £1.1 billion is insufficient for a UK foundry.
The funds could bolster domestic chip designers like Arm Holdings, based in Cambridge but listed in New York. The government vaguely mentioned a "strategic industry partnership" with Arm. A £400 million procurement opportunity for UK chip makers is the strongest component, but industry experts note much of this was previously announced. Mark Boost, CEO of Civo, warned that without deliberate contract structuring, the money might flow to established overseas vendors, resulting in "British-branded infrastructure on somebody else's silicon."
AI Skills and Adoption
The government committed £20 million to map how AI changes entry-level work and provide business advice for role redesign. A "bridge AI" scheme will fund UK companies to buy domestic AI products, and the "tech town" programme, pioneered in Barnsley, will expand. Bespoke AI adoption plans were published for advanced manufacturing and creative industries.
Bouke Klein Teeselink from King's College London noted that few people use AI tools to full potential, leaving productivity gains untapped. However, he argued that private sector adoption will likely be more efficient than government-backed programmes, which may move too slowly.
Defence and US Chip Investments
Britain's chief of defence staff, Sir Richard Knighton, announced the Rapid AI Delivery Taskforce (RAID) to develop AI models for defence, while maintaining that "humans, not machines, are accountable for decisions."
AMD and Nebius announced UK investments. AMD said it will invest "up to £2bn" to accelerate AI innovation with the University of Cambridge and Imperial College. Nebius committed "approximately £1.7bn" for AI infrastructure, which appears to involve Nvidia chips.
Ordering Apple and Google to Tackle Nudity
The government demanded that big tech providers, including Apple and Google, detect and block nude images for children through technical solutions or built-in features. Non-compliance could lead to criminal liabilities and fines. This is a more stringent demand than earlier requests to remove non-consensual nudes.
Currently, companies like Google and Meta use digital watermarking to flag terrorist content and child sexual abuse material, relying on user reports. Detecting and blocking nude images for children would likely require more invasive measures: age verification for all users, potentially involving identity verification, and extensive scanning of user-shared content. This could force companies to store large amounts of user data, risking leaks or government subpoenas.
Privacy-focused companies like Signal and Mullvad have raised alarms, warning of a "dystopian combination of age verification and content scanning" that could enable "mass censorship capabilities." They argue that internet privacy is essential for activist groups organizing against authoritarian regimes.
Looming Social Media Action
The government is expected to announce a ban on high-risk social media apps for under-16s next week. The specific apps are unknown, but Australia's blanket ban covers Facebook, Instagram, X, TikTok, and Snapchat. Features like disappearing messages on "safe" apps may also face age restrictions.
Age verification remains a key issue. Questions include whether tech companies should request and retain government-issued IDs, how to handle security, and what to do about children without IDs. Meta suggests app stores (run by Google and Apple) should handle age verification. Google is developing "age assurance" measures, potentially default guardrails for unknown users and occasional ID requests. AI-powered facial age estimation is another option, though children in Britain have reportedly circumvented it by painting on moustaches.
