PolicyLaw, regulation & governance
UK Chancellor John Healey unveils AI security package ahead of budget, including a biosecurity lab co-funded by OpenAI and Anthropic
Ahead of the 28 October budget, Chancellor John Healey announced an AI package reported at more than £600m. It centres on an AI biosecurity lab part-funded by OpenAI and Anthropic and adds £120m for the AI Security Institute and cyber-defence spending.
From the The Guardian · AI article of October 24, 2026 and cited sources: City AM puts the package at more than £600m, while the Guardian describes it only as hundreds of millions. [1] [2]
SOURCES[1] John Healey pledges to use budget to harness power of AI… theguardian.com[2] City AM: Healey vows AI will not 'leave workers behind' in… cityam.com
AT Pulse research deskPublished · Updated · 5 min read · AI-assisted analysisAI-assisted, not human-edited. Methodology Corrections Who owns this site
- What happened
- City AM puts the package at more than £600m, while the Guardian describes it only as hundreds of millions. [1] [2]
- Original source
- The Guardian · AI article, independent perspective
- Brief by
- AT Pulse research desk AI-assisted; publisher and contact on the desk page
- When
- October 24, 2026 article published
- Covered
- Oct 24 edition
- Evidence
- 11 sources across 11 sites
- 1John Healey pledges to use budget to harness power of AI…theguardian.com
- 2City AM: Healey vows AI will not 'leave workers behind' in…cityam.com
- 3LBC: UK to build £300m AI 'biosecurity lab'lbc.co.uk
- 4aiera.blog: Kanishka Narayan AI safety warningaiera.blog
- 5ARY News: Government websites go dark after AI-discovered…arynews.tv
- 6The Business Journal: AI discovery of PHP vulnerabilitiesthebusinessjournal.com
Reach: not measured yet
City AM puts the package at more than £600m, while the Guardian describes it only as hundreds of millions. The figures come from Treasury briefings rather than published budget documents. Reported items include £120m for the AI Security Institute, a £60m National Cyber Shield, a £60m top-up to the Sovereign AI Fund and £10m a year to study AI's effect on jobs. [1] [2]
The largest item is an AI biosecurity lab worth up to £300m. It would use AI to detect biological threats and develop medical countermeasures, and Google DeepMind and Basecamp Research are named as partners. Up to £200m is linked to OpenAI and Anthropic, but outlets disagree on whether that means cash or model credits. Credits would be worth much less than cash. [2] [3] [8]
The extra money for the AI Security Institute follows AI Minister Kanishka Narayan's plan for the institute to evaluate frontier models before release. Critics question whether that testing will be enforceable. Opposition figures called the wider package a list of projects without a clear strategy. [3] [4] [8]
The reported funding would increase the UK's capacity to test frontier models before release and to use AI against biological threats.
Read the full assessment
One possible but unconfirmed result is that pre-release evaluation becomes an expected step for AI companies launching in the UK.
Executive brief
The biggest single item in Chancellor John Healey's AI package is a new AI Biosecurity Lab worth up to £300m. Up to £200m of that is reported to come from OpenAI and Anthropic, though outlets disagree on whether it is cash or model credits (City AM; LBC). The lab sits inside a package of more than £600m announced ahead of the 28 October budget. The package also gives £120m to the AI Security Institute, £10m a year to study AI's effect on jobs, and money for cyber defence (The Guardian). It was announced in a week when an AI-discovered PHP flaw took websites offline across Europe.
What changed and event timeline
Narayan warns a breach is "likely, if not inevitable"
In his first cabinet-level speech, AI Minister Kanishka Narayan announced stronger cyber legislation and said the AI Security Institute would evaluate frontier models before release ().
AI-found PHP flaws force sites offline
The PHP Foundation issued emergency patches for two remote-code-execution bugs found with AI tools. Government portals in the Netherlands, France and Finland went offline (;).
Households told to prepare for emergencies
Defence Secretary Wes Streeting launched "Be Ready", which advises households to keep 72 hours of supplies for weather, cyber or other emergencies ().
Healey announces a £600m+ AI package
The package includes the biosecurity lab, £120m for the AI Security Institute, a £60m National Cyber Shield and £60m more for the Sovereign AI Fund ().
Budget scheduled
This is Healey's first budget as chancellor in Andy Burnham's government ().
Capabilities and access
- No model release. This is a funding and institution-building package.
- Biosecurity lab: £165m goes to a National AI Biosecurity Initiative, including £100m for a lab that will use AI to detect biological threats and develop medical countermeasures.
- Model access: no specific OpenAI or Anthropic models or versions are named.
Read the full section
- No model release. This is a funding and institution-building package.
- Biosecurity lab: £165m goes to a National AI Biosecurity Initiative, including £100m for a lab that will use AI to detect biological threats and develop medical countermeasures. Google DeepMind and Basecamp Research are named as partners (City AM).
- Model access: no specific OpenAI or Anthropic models or versions are named.
- Other items: AI chip design in Edinburgh, a Centre of Excellence in Manchester (The Guardian), and a £70m NERV centre to detect cyber threats (BritBrief).
Technical analysis for researchers and developers
- No technical design is public yet.
- The extra £120m for the AI Security Institute follows Narayan's plan for it to test frontier models before release.
- The PHP incident shows AI-driven bug-finding at scale. The bugs are CVE-2026-109618 and CVE-2026-109619, rated CVSS 9.2 (The Business Journal).
Read the full section
- No technical design is public yet. The reviewed sources do not describe the lab's architecture, evaluation methods, data-access rules or how its work will be published.
- Pre-release testing. The extra £120m for the AI Security Institute follows Narayan's plan for it to test frontier models before release. Critics question whether that testing will be enforceable (aiera.blog).
- The PHP incident shows AI-driven bug-finding at scale. Wiz reportedly used its Atlas tool alongside Gemini 4 Argon and GPT-5.6 Sol to find flaws that had sat undetected for more than a decade (ARY News). The bugs are CVE-2026-109618 and CVE-2026-109619, rated CVSS 9.2 (The Business Journal).
Claims and evidence
- £600m+ total: reported by City AM. The Guardian says only "hundreds of millions".
- Lab funding conflict: City AM describes up to £200m in model credits. LBC and BritBrief say £200m "from" OpenAI and Anthropic.
- UK had the G7's strongest growth in the first half, helped by AI investment: attributed by the Guardian to official figures, but not checked against those figures here.
Read the full section
- £600m+ total: reported by City AM. The Guardian says only "hundreds of millions". All figures come from Treasury briefings, and no official gov.uk document appeared in the reviewed sources.
- Lab funding conflict: City AM describes up to £200m in model credits. LBC and BritBrief say £200m "from" OpenAI and Anthropic. Credits are worth much less than cash, so this matters.
- UK had the G7's strongest growth in the first half, helped by AI investment: attributed by the Guardian to official figures, but not checked against those figures here.
- Barclays cutting 5,000 jobs: reported by the Guardian and City AM. No other confirmation found.
Context and prior work
- The AI Security Institute already works with frontier labs to test models for biological risks (Parallel Parliament).
- The UK Biological Security Strategy aims to build national resilience to biological risks by 2030 (HSToday).
- Healey had already called AI the "greatest innovation of our time" and promised workers would not be left behind (Irish News).
Limitations, safety and contested findings
- Critics called the package a "shopping list of projects" with no clear strategy (BritBrief).
- Shadow AI Minister Ben Spencer questioned whether UK rules will align with the EU's and urged removing business barriers instead of directing AI development (LBC).
- FCA chief Nikhil Rathi warned that new technologies pose growing risks to markets (City AM).
Read the full section
- Critics called the package a "shopping list of projects" with no clear strategy (BritBrief).
- Shadow AI Minister Ben Spencer questioned whether UK rules will align with the EU's and urged removing business barriers instead of directing AI development (LBC).
- FCA chief Nikhil Rathi warned that new technologies pose growing risks to markets (City AM).
- Two frontier labs funding a lab that may help oversee their own models' biological risks raises a potential conflict of interest. The reviewed sources do not describe any governance safeguards.
- Fiscal pressure is real: oil topped $117 a barrel on 23 October (The Guardian).
Business and practitioner implications
- Cloudflare reports early attacks already emerging (The Business Journal). Expect ministerial "direction powers" under the Cyber Security and Resilience Bill (aiera.blog).
- UK AI firms: possible openings through the £60m Sovereign AI Fund top-up, £30m for local AI partnerships and a £15m health-AI competition. techUK welcomed the package (City AM).
- Frontier labs: pre-release evaluation by the AI Security Institute may become a condition for launching in the UK.
Read the full section
- Security teams: patch PHP now. Cloudflare reports early attacks already emerging (The Business Journal). Expect ministerial "direction powers" under the Cyber Security and Resilience Bill (aiera.blog).
- UK AI firms: possible openings through the £60m Sovereign AI Fund top-up, £30m for local AI partnerships and a £15m health-AI competition. techUK welcomed the package (City AM).
- Frontier labs: pre-release evaluation by the AI Security Institute may become a condition for launching in the UK.
- Before planning around any figure: wait for the 28 October budget documents.
The source trail.
Sources (11)
John Healey pledges to use budget to harness power of AI to bolster Britain’s security
Article text retrieved; extracted text may omit tables or interactive elements.
theguardian.com