Blog

The press release issued by the Department for Science, Innovation and Technology on 12 June 2026.

The Government Cyber Action Plan aims to boost cyber resilience across the UK public sector by using emerging technologies to manage risk. The Government Cyber Coordination Centre (GC3) – a partnership between the NCSC and the Department for Science, Innovation and Technology – is leading this work, exploring how frontier AI can be applied safely to cyber defence across government.

From frontier models to front-line impact 

We know AI is disrupting the cyber threat landscape. Recently released frontier AI systems such as Claude Mythos and GPT-5.5 brought a step-change in cyber capabilities, and the UK AI Security Institute (AISI)’s evaluations show these models getting better at cyber tasks very quickly.  

However, evaluation in synthetic environments gives a limited understanding of real-world use. A high score on a benchmark does not necessarily translate into finding and fixing real vulnerabilities.

What we did 

The Government Cyber Coordination Centre led a weekly, in-person series of hackathons which used frontier AI to scan public code repositories across government. Working closely with specialists from the AISI and NCSC, our goal was to find and mitigate previously unidentified vulnerabilities before they could be exploited. Rather than mandate a single approach, we gave teams model access and let them build their own tooling, noticing what worked each week and building on the best approaches. 

The UK Government encourages new source code to be open by default, with specific and justified exceptions. In practice, that creates a degree of shared visibility that attackers can also exploit. However, this openness also limits duplication and leads to cleaner, more easily maintained code. 

Code published in the open has also already passed extensive prepublication scrutiny, meaning it can be shared with frontier model providers with minimal additional review. This means that government departments can deploy new capabilities quickly and with confidence.

An adversarial chain that challenges itself. One team ran each public repo through a six-stage AI agent pipeline: triage, validator, auditor, tracer, judge, summary. Each stage reads and challenges the last. In one case, the agent downgraded a finding once it established that a backup mechanism was in place. The pipeline was agentic, but the escalation was manual. This means a member of the team checked every line, re-verified exposure, and handled false positives.

Deterministic scanners feeding a model. Another team ran traditional scanning tools first (including Gitleaks, Trivy, Semgrep and Hadolint) to generate a ranked findings document. Three model stages were then layered on top: a discovery stage that treated the scanner output as leads and read the source against OWASP and CWE frameworks, a chain-investigation stage that composed individual findings into attack paths via per-chain sub-agents, and a triage stage that confirmed the finding viability.

Codifying a multi-service audit into reusable skills. Another department developed five domain-specific Claude Skills. The Skills distil an organisation wide audit across hundreds of services into something repeatable. Skills enabled a reusable, scoped, and consistent approach across every repository and operator.

What we found 

Participants identified 407 findings in total, including critical weaknesses exposing services to authentication bypass, data exposure and remote code execution. Some were already understood and mitigated by compensating controls while others were previously unknown. All critical weaknesses have been remediated, and no evidence of exploitation was identified for any finding. 

AI models traced vulnerabilities across service boundaries, which traditional scanners can’t do, and linked business logic with technical detail. Departments prioritised validation and remediation through existing frameworks, patching critical and high-risk issues assessed as exploitable. 

It cost us £13,000 in tokens to find these weaknesses, working across nine government organisations for the month.

Identifying Critical vulnerabilities: One notable finding affected legacy GitHub Actions in a repository supporting a key government digital service. The issue allowed an external user to trigger a workflow chain by posting a specially structured comment on an open pull request. This bypassed the usual protections for pull requests from unknown contributors because the workflow was triggered by a comment, not by the pull request itself. 

The impact was arbitrary remote code execution on the GitHub Actions runner. The workflow took content from the comment, passed it into deployment parameters, and used it in an environment substitution step that executed during the workflow. By placing executable content in the comment field, an external user could cause their input to run on the GitHub runner. 

This created a route for malicious actors to potentially extract secrets and tokens available to the workflow, including the GitHub token used by the automation. With that level of access, the issue could support wider repository compromise, including manipulating pull requests, approving workflow activity, altering trusted contributor status, and exploit further secrets available to the automation environment.

What we learnt 

Across teams, the common thread was structure. Models were used as components, using Skills, running in parallel across repositories, and a human expert kept in the loop on anything that mattered. We learnt that: 

• Architecture matters the most. The strongest results came from using frontier models as tightly scoped components inside a structured pipeline. Breaking traditional vulnerability management workflows into discrete, task-specific harnesses let teams scale while controlling false positives and hallucination. 
• The model matters less than how it’s used. AISI’s research, borne out here, shows that with the right architecture and task design many near-frontier and frontier models perform comparably at scanning code. The best findings still lean heavily on human expertise in breaking the problem down and identifying wider context. 
• Triage is essential. Agents generate candidate findings far faster than humans can validate them. Poorly scoped runs burn tokens on low-value targets; weak review dumps the load onto stretched security teams. Careful upfront scoping and structured internal filtering of low-confidence findings kept human review focused. As in traditional vulnerability management, it’s not how many issues are found, but whether triage points limited resource where it matters. 
• Finding isn’t the same as fixing. Findings still had to enter the patch pipeline for remediation. AI shows promise here too, but today prioritisation, review and patch-generation all must integrate without overwhelming human-centred processes.

What next 

GC3 will kick off a second phase of this pilot, with more departments, additional models, and an extension from public code to closed-source estates. Identifying vulnerabilities early on, raising the consistency of defensive practice, and helping departments share on proven techniques is how we put the Government Cyber Action Plan into practice.  

AISI and NCSC’s involvement will also deepen as we continue to evaluate AI as a tool for cyber defence in applied settings, closing the gap between a theoretical benchmark and a real reduction in risk. 

This pilot was a test of how government can adopt new capabilities responsibly, learn quickly, and share what works.

The press release issued by the Department for Science, Innovation and Technology on 12 June 2026.

More than £6 billion of new investment and around 8,000 new jobs have been announced this week, as companies from around the world choose to build, hire and scale here.

More than £6 billion of new investment and around 8,000 new jobs have been announced this week, as companies from around the world choose to build, hire and scale here – reinforcing Britain’s position as a global hub for AI.  

From AMD’s £2 billion commitment for next generation AI compute, to Nebius investing £1.7 billion in new infrastructure, to homegrown companies like Oxford Quantum Circuits securing record funding – this is Britain’s AI leadership turning into real jobs, investment and opportunities across the country. 

These deals span the full breadth of the AI economy – from chips and cloud infrastructure to autonomous vehicles and open-source development – showing the UK isn’t just keeping pace in the global AI race but helping to shape its direction.  

The momentum has been matched by government action throughout the week to strengthen the UK’s tech sector – backing the skills, infrastructure and innovation needed to unlock growth. That includes new support for young people to seize the opportunities of AI, the first-ever AI Adoption Summit, a landmark £1.1 billion AI Hardware Plan, plus new backing for open-source AI builders, and a data centre design challenge – so Britain builds with taste.  

Technology Secretary Liz Kendall said: 

Britain is seizing the opportunities of tech and AI to create jobs, improve lives and grow businesses.

Companies from across the globe are choosing to invest here and hire here, bringing billions of pounds and thousands of new jobs with them.

And this Government is backing them with our £1.1billion hardware plan and our investment in skills and training.

We are rebuilding Britain for the modern age and creating a future that works for all.

A range of major investments have been announced, totalling over £6 billion of investment and around 8,000 new jobs – from global leaders and frontier companies choosing to grow here. This includes:

• AMD commits up to £2 billion of investment over five years to Accelerate AI Innovation and Research in the United Kingdom.  
• Nebius, the AI cloud company, announced it is investing approximately £1.7 billion to build out capacity in the UK with three new deployments of advanced NVIDIA compute, as the company continues to expand its commercial and AI R&D hub in London. 
• Amazon opened a new fulfilment centre in Northampton and announced plans for a second major site in Kettering, committing more than £1 billion and up to 4,000 jobs in a single county as part of its planned £40 billion UK investment. 
• British unicorn Ark has announced an investment of £807m in the expansion of its Longcross Park campus. The expansion supports the deployment of leading AI cloud provider Nebius, which is also increasing its footprint in the UK as well as an additional data centre facility with 36MW of future capacity. 
• Eros Innovation is investing £265m and establishing a sovereign British Cultural AI capability, supporting 3,000+ jobs across 15 productions over five-years (2 films shooting in the UK in 2026), licensing its $1.7bn cultural dataset to its UK operation and launching an AI Studio to develop Large Cultural Models trained on British creative heritage and governed by British law. 
• Quantum computing scale-up Oxford Quantum Circuits (OQC) securing a £260 million investment – the largest quantum funding round in the UK, backed by the British Business Bank.  
• Silicon Valley investors Playground Global are launching a new fund backed by up to £150 million from the British Business Bank – the largest fund investment the bank has ever made – to invest in UK-based hardware companies and help them scale.  
• Arlequin AI investing up to £45 million in the UK over the next 5 years with capital deployed across local hiring, UK R&D, and sovereign deployment capability for government and critical national infrastructure clients. 
• Midlands Mindforge has now entered its active investing stage, thanks to the support of the Mayors of the East and West Midlands, UK Research and Innovation (UKRI), the Rigby Group and the Department for Science, Innovation and Technology (DSIT) via the Invest in UK University R&D Midlands Campaign. This will help to unlock an initial £30m of capital into the region making its first round of investments into spinout companies. 
• Cosine announced the formation of a coalition of major UK institutions including, BAE Systems, BT, Lloyds Banking Group, NatWest, PwC and Leonardo UK, to co-design Lumen Sovereign, Britain’s first fully sovereign frontier AI model. Backed by the Government’s Sovereign AI Fund, Lumen Sovereign will be trained entirely on UK soil using Isambard-AI. 
• AI coding startup Cursor announced plans to open its European headquarters in London. 
• Fynd, the AI-native unified commerce platform serving brands and retailers across India, the Middle East, Africa and Southeast Asia, has opened its first UK office in London. This investment is projected to create 70 jobs across London, Manchester and the wider UK by 2028. 
• General Intuition — the frontier research lab dedicated to foundation models for spatial and temporal reasoning — has opened a UK office in London King’s Cross. The lab will invest some of its recent $133M seed round into expanding its London-based research team, drawing on the UK’s strong talent base. 
• Legora, the AI platform for legal professionals, is expanding its European footprint with a dedicated engineering hub in London in a major vote of confidence in the UK’s AI capabilities.  
• Multiverse announced the opening of a new technology hub in Edinburgh and plans to create 200 jobs in the next year across the new office and its London headquarters. 
• German AI unicorn n8n will expand its investment into the UK by delivering up to 200 high-skilled jobs over the next three years, a strong endorsement of the UK’s AI talent ecosystem. 
• PhysicsX secured a $300 million Series C investment, taking its valuation to approximately $2.4 billion. The funding will accelerate PhysicsX’s global expansion, platform development, and frontier physics AI research. 
• Reflection, the US-based open-source AI lab founded by former Google DeepMind researchers Misha Laskin and Ioannis Antonoglou, is expanding its UK footprint with plans to hire more than 100 highly skilled employees within the next 12 months, growing to over 1,000 roles within three years. 
• US tech company Replit will be opening up an office in London this year.