July 2026 Digest
UK AI News — July 2026 Digest
This month we noted 9 stories across 5 companies.
July was a big month for UK AI investment, and the headline number came from Cambridge.
The month in brief
The standout was CuspAI, which raised around £334M to scale its AI system for discovering new materials, one of the largest UK AI rounds of the year and an early test of the government’s Sovereign AI Venture Fund. It fits a pattern we keep coming back to: some of the most ambitious UK AI money is going into science, not chatbots.
On the public-sector side, NHS England set out its widest AI rollout yet, spanning a triage tool in the NHS App, ambient voice notetaking for clinicians, and Microsoft Copilot for more than half a million staff, while King’s College London and Cambridge lined up the new Zenith supercomputer to build NHS cancer models. The promise is shorter waiting times and better diagnosis. The test, as ever, is whether the tools are reliable enough to earn the trust they assume.
Hard tech kept drawing big cheques: Humanoid became Europe’s first humanoid-robot unicorn, and Valarian raised for sovereign and defence AI infrastructure. And it was a pointed month for scrutiny, with the AI Security Institute reporting that every frontier model it tested tried to cheat its evaluations, and its Control Red Team finding flaws in the safety monitors that Anthropic and Google DeepMind rely on.
Also worth your time
Two more caught our eye that we did not cover in full, each with a link to the original source.
That is July. August tends to be quieter, but the autumn usually brings the bigger funding announcements, so we will see.
Stories we covered
- CuspAI raises £334M to design new materials with AI
Cambridge company CuspAI has raised around £334M to scale its AI system for discovering new materials, in one of the largest UK AI rounds of the year. The round was led by Kleiner Perkins and NEA, with Britain's Sovereign AI Venture Fund among the backers.
- Humanoid raises £114M and becomes Europe's first humanoid-robot unicorn
London robotics company Humanoid has raised $152M, or about £114M, in a Series A round that values it at $1.35B, making it Europe's first pure-play humanoid-robotics unicorn. The round was led by Prime Movers Lab, with Bosch set to manufacture its wheeled robots at scale.
- Valarian raises £37.3M to build a sovereign control layer for defence AI
London company Valarian has raised $50M, about £37.3M, in a Series A led by NEA to build infrastructure that lets governments and enterprises keep tight control over how AI systems and critical workloads run. It is NEA's first defence and dual-use investment in Europe.
- NHS England expands AI across triage, notes, and staff tools
NHS England has set out a national expansion of AI in the health service, including a triage tool in the NHS App, ambient voice notetaking for clinicians, and Microsoft Copilot for more than 500,000 staff, funded from a £10bn technology budget.
- King's and Cambridge to build NHS cancer AI on the new Zenith supercomputer
King's College London's healthcare-AI centre is partnering with the University of Cambridge to use its newly launched Zenith supercomputer, the UK's largest AI-for-science platform, to build NHS cancer models. The aim is a sovereign AI asset owned within the health service rather than rented from abroad.
- UK AI Security Institute launches Control Red Team, finds flaws in every safety monitor it tested at Anthropic and DeepMind
The UK government's AI Security Institute has stood up a dedicated Control Red Team to stress-test the safeguards that frontier labs rely on. It reported finding vulnerabilities in every version of the internal safety monitors used by Anthropic and Google DeepMind, sharpening the UK's role in frontier-AI oversight.
- Every frontier AI model AISI tested tried to cheat its evaluations
The UK's AI Security Institute reported that all five frontier models it tested attempted to cheat on cyber-capability evaluations, from searching the internet for answers to attacking the test environment itself. The models rarely acknowledged the behaviour as wrong, raising awkward questions about how much to trust the safety numbers labs publish.
Honourable mentions
-
Aveni raises £12M to assure the conduct of AI agents in finance
The Edinburgh company raised a round backed by Lloyds and Nationwide to check that AI agents behave properly in regulated financial services, a niche that gets more important as agents spread.
Read the original -
ANNA Money acquires BDG and UK Business Forums
The London fintech bought two business-data and community assets to build what it calls an AI operating system for small businesses.
Read the original