Whatever your view on the Fable decision - it has bought both nation states and individual companies a bit more time to test their cyber vulnerabilities - hopefully they will use it! Some advice from DSIT on this below for businesses in the UK: gov.uk/government/publicatio…

A few words on the Sovereign AI debate, having built several LLMs in Meta while in the UK and now working as a UK based startup: 1. Lots of people are trying to do the right thing to make the UK a better place to start AI companies. Time lags until the benefit show, but you should judge on the intent now. I support the direction of travel! 2. DeepMind has been enormously beneficial for the UK, but it has muddied the waters for a sovereign LLM company to emerge as (until recently) the Government continued to celebrate it as a British achievement / push it as a national champion. 3. Similarly, people are now celebrating recent US investment in King’s Cross, while also wanting more UK sovereignty. Clearly some income effects here, but I would worry about the substitution effects too. AI is not like other types of foreign investment. 4. The relevant talent nexuses in UK that could develop a competitive foundation model are from GDM and old Meta AI GenAI. Also some folks from smaller groups, ex Conjecture, Stability. The talent is still there, although a lot was snapped up by US FM companies in the past year. I personally think it’s not too difficult to develop new talent either from UK universities, but you probably need an ex GDM or Meta core (Gemini or Llama). Or if not: show evidence first (technical reports) before claiming you can do it. 5. Building an LLM is very different from doing regular AI research - skillset is different. Former is closer to engineering; long hours, often unsexy work. Important to distinguish between these two types of talent in the UK ecosystem; arguably too much focus on the latter / ideas guys. 6. On research - DeepSeek R1 post-train cost $300k . Yes, they also needed an ablation budget and to train a base model, invest in infra and talent - and yes the cost of an R1 moment is increasing year on year - but the idea that you need $1bn plus immediately to show results is complete FUD. You need billions to scale, not to validate new directions. 7. In my experience, every failed LLM effort (from model results perspective) I witnessed in the past came from a combination of poor leadership, politics, unclear vision, and premature scaling. Good efforts usually started from small teams who had worked with each other for a long time, had shared thesis, and scaled progressively in bite-sized pieces. Some recent lessons here for neolabs as well. 8. Things take time. Eg we’ve spent ~12 months mostly on internal infra just to get into the position to be able to make big swings. It’s important to nurture new companies through the initial phase. Expectation management is also crucial. I think expecting new UK companies to have single big bang releases is very dangerous; sort of like overwatering a plant. The correct release pattern is “decent”. “decent”, “decent”, “quite good actually”, “holy shit”. 9. Please don’t allow politicians or journalists to kill recent or upcoming AI investment efforts. We will need way more - at the price of potential inefficiency in places - as AI is existential for the country. Ambitious projects are usually incredibly fragile in the early stages; look after them! 10. Mythos is a good triggering moment, but what’s coming will make it look like a toy, so it’s worth building for what’s coming in 5 years time - not a current generation model. Very proud to be building in the UK - more to share on that soon - alongside many other great early stage AI companies! 🇬🇧

Another reminder of just how critical sovereign AI capabilities are. As an aside, @UKSovereignAI is hiring: sovereignai.gov.uk/about

I get asked often: "AISI's evals show models getting rapidly better at cyber. But does that translate into the real-world capability?" So with government partners, we ran AI systems across real government code-bases. 407 findings, incl. multiple critical vulnerabilities. 🧵

I was delighted to speak at the AI Summit during @LDNTechWeek today. AI has the capacity and capability to transform warfare. If we don’t make progress and recognise the profound effect AI will have on the business of war, we will lose and we won’t deter.   Read my speech 👇

In the midst of an intense week in UK tech ( and globally ) it’s worth remembering that the Department for Technology in the UK (DSIT) didn’t exist just 3 years ago. At the start of ‘23, we didn’t have anyone in Cabinet whose sole role was to help Govt navigate what is now one of the most important opportunities and challenges facing the country. Perhaps because of the Dept’s leadership ( it’s not clear that @leicesterliz or @KanishkaNarayan have slept this week! ) and it’s willingness to experiment with new approaches like ARIA, Sovereign AI and AISI, it’s been one of the most impactful and creative arms of Govt in recent years. At a much needed time. Yes, the areas for improvement are endless. But institutional reform is probably the area of biggest delta in votes to impact ( no votes in it, huge impact if it works ). Credit to @RishiSunak and everyone involved in creating this Dept. I think everyone involved in UK tech week hopes the trajectory continues.

We're doubling compute access for the UK's most ambitious AI startups. The response since launch has been remarkable. In just three months, we have received more than 200 applications requesting over 67 million GPU hours. Founders here are building frontier AI models, scientific discovery platforms, next-generation compute infrastructure, cybersecurity technologies, drug discovery tools, and other innovations with the potential to shape the next decade of economic growth and national capability. If Britain wants to be an AI maker, not an AI taker, we need to back our most promising startups with the compute they need to build at the frontier. Today's announcement at @LDNTechWeek does exactly that.

A London startup called Cosine is building what it calls Britain's first sovereign frontier AI model. The model is called Lumen Sovereign. An AI you can run entirely inside your own walls.You dont' need external data transfer,foreign managed servers and code leaving the building.For banks, hospitals and defence primes,thats the only version that's legally allowed in the room. The UK government chose @CosineAI as a first partner in its £500 million Sovereign AI programme after the company's models outperformed OpenAI, Anthropic, Mistral and DeepSeek on independent coding benchmarks for two straight years. They were handed 500,000 GPU hours on Isambard AI, one of the most powerful supercomputers in Europe. Cosine is already working across UK defence primes and critical infrastructure including organisations tied to the UK's nuclear deterrent. Congratulations to @yangli_ , @AlistairPullen , @SamOfStenner n all the team members of Cosine

British startups to “scale here” and “stay here”, as UK PM unveils £400M chip plan tech.eu/2026/06/08/british-s…

As well as incredible examples of entrepreneurship and innovation, it's worth remembering that the US Government was the first customer of both SpaceX and Palantir ( and in Palantir's case, one of the first investors too ). Both have been hugely positive for the US economy and their geopolitical position. There will be many positive announcements this week from my colleagues at DSIT on backing British AI start-ups, so it's worth remembering the size of the prize if we get it right. Government procurement isn't about picking winners, but supporting innovation and providing better public services. But back when SpaceX and Palantir were starting up and getting US Govt support - the UK couldn't have procured from a local start-up ( EU frameworks would have prevented them ) and even if we wanted too, the Govt didn't prioritise working with new technologies - but went with 'safe bets' with legacy companies for institutions like the NHS and Post Office....look where that got us. Thankfully, if belatedly, with the introduction of the BBB, NSSIF & the Procurement Act under the Conservatives, and expansion of the BBB and set-up of SovAI under this Government, I'm much more positive about our ability to start and scale similarly important, international winners going forward.

🚨New analysis from us @BCG CfG 🚨 Ahead of today's UK AI Adoption Summit, we have looked at the potential productivity & economic gains from AI adoption in the UK. There is significant potential, up to £1 trillion added to the UK economy over 10 years. But it depends on the right kind of adoption. This means deep adoption & integration of more sophisticated AI tools, not just bolting on off the shelf tools. 🧵on our key findings & new insights (from part 1 of our work, with part 2 to follow later in summer) 1/

Former (Conservative) Prime Minister Rishi Sunak praises (Labour) Government’s awesome SovAI unit. UK’s AI work remains one of the few ongoing areas of senior cross party agreement in British politics.

We’ve had an astonishing 338 British AI companies apply for our first ‘strategic assets’ programme at SovAI. We aim to support a handful with multi-£M offers to build essential infrastructure here in areas like advanced wet-labs. The team are working through it all - but as they say - we’re going to need a bigger boat!

A team at Oxford built a search engine for every drug the NHS prescribes, and it has quietly saved the health service millions. It's called OpenPrescribing. The NHS publishes its full prescribing dataset every month. It's 700 million rows of raw numbers nobody could actually read. So Oxford built a tool that turns it into live charts in seconds. You type a drug name. It shows you which practices over-prescribe it, which regions are slow to follow new guidelines, and where the money is being wasted. → Search any drug across any GP practice in England → Find safety and cost outliers instantly → 70 ready-made quality measures → Updates monthly, automatically → Free, open source, MIT licensed 20,000 people use it every month. Doctors. Researchers. Journalists. Public data that sat unreadable for years is now one search away. openprescribing.net/analyse/