PauseAI UKUK
← All briefings

By Annabelle Collins ·

    • Lords: Cyber Bill
    • Anthropic files S-1
    • Budget day
    • Nordic AI meeting

AI News Weekly Briefing - 23rd-30th October

Key stories this week

  • Monday: Anthropic files to go public. Anthropic, one of the big two “frontier” AI companies, filed its IPO to list on the US stock market.1
    Its prospectus warns that AI models, including its own, have shown “self-preserving behaviors, including attempts to resist shutdown”, and says it now uses its AI to help build the next successor model.1
  • Wednesday: Budget released. £700m for AI Roughly £480m goes on AI security, including a new AI Biosecurity Lab and £120m more for the AI Security Institute.2,3

Budget Breakdown

This week Chancellor John Healey released his first budget, committing ~£700M funding for AI.3

Most of it pays for defences against AI-enabled threats such as engineered pathogens or rogue AI agents.2,3

“Nowhere is the necessary role for the state clearer than in Artificial Intelligence.”4

John Healey, Budget speech

Security & safety£479m£695m in total
Sovereignty & industry£138m
£78mJobs & the public
  • Security & safety
  • Sovereignty & industry
  • Jobs & the public

£695m in total

What it’s for
National AI Biosecurity Initiative£100m bio-threat lab + £65m surveillance
165
£100m bio-threat lab + £65m surveillance
AI Security Institute (top-up)State lab testing frontier AI for dangers
120
State lab testing frontier AI for dangers
NERV vulnerability centreAI hunts weak spots in key infrastructure
70
AI hunts weak spots in key infrastructure
National Cyber ShieldAI agents defending UK networks
60
AI agents defending UK networks
Sovereign AI Fund (top-up)Invests directly in British AI firms
60
Invests directly in British AI firms
Sovereign AI R&D procurementState buys UK-built AI (health, security)
60
State buys UK-built AI (health, security)
Agentic security unit ★Watches for rogue AI agents; responds
40
Watches for rogue AI agents; responds
Local AI partnerships (mayors)Mayor-led AI adoption in the regions
30
Mayor-led AI adoption in the regions
Public-services cyber ★Protects hospitals, schools and councils
24
Protects hospitals, schools and councils
AI Economics Institute (3 yrs)Studies AI’s effect on growth and jobs
23
Studies AI’s effect on growth and jobs
Advanced Chip Design AcademyTrains British chip designers
18
Trains British chip designers
AI Challenge PrizeRewards AI that fixes everyday problems
15
Rewards AI that fixes everyday problems
AI & jobs research (per year)Tracks how AI reshapes work
10
Tracks how AI reshapes work
£0m£50m£100m£150m
£ millions. ★ = first reported on Budget day (City A.M.). Model credits from OpenAI and Anthropic (up to £200m) not shown.2,3

Anthropic’s IPO filing

Anthropic, the company making the Claude AI models and one of the two leading "frontier" AI developers, is going public. On Monday it released it’s S1.1
This is an IPO prospectus and and the company is legally liable if it leaves out or misstates material risks.

“Our ability to maintain our position at the frontier of AI depends in part on our use of AI systems to accelerate our research processes, a recursive dynamic we refer to as “closing the loop,” in which successive versions of our models assist in building the next, with the goal of compounding capability gains over time.”1

- Anthropic S1 filing

“They gave an account in here about how the company's models have exhibited what they call ‘unexpected, strategic, or self-preserving behaviors’, including attempts to resist shutdown, conceal or manipulate information, or generate outputs that could be interpreted as coercive, deceptive, or manipulative.”9

- Casey Newton, Machine Gods 28th Oct

In Westminster this week

Mon 26 Oct - Lords - Cyber Security and Resilience Bill, report stage

The government has said it will use this Bill to bring in its first protections against AI risks.5 These include AI “loss of control”, where autonomous AI systems evade human oversight.5 The report stage is the Lords’ last detailed chance to change the Bill.

What changed on Monday? Now …

  • Within a year, ministers must review whether the AI Security Institute (AISI) should get legal powers, such as auditing frontier AI before release.6,7
  • If an AI model causes an incident, the duty to report stays with the organisation that was hit, such as a hospital or utility, not the model’s developer.6,7
  • New “loss of control” powers. If an AI agent escapes human control and causes a cyber incident, regulations can give ministers powers to send in inspectors after that incident. Those investigators could enter premises, run tests, demand information held abroad and fine firms.6,7

The third reading will be on Monday 2 Nov.7

26th Oct - Commons - Debate on Emerging risks of advanced AI debate - POSTPONED

This was due to be the first Commons debate on regulating advanced AI, led by Plaid Cymru MP Ben Lake.8 It was postponed for lack of time.
A new date may be announced in the Business Statement on Thursday 5 November.

International Cooperation

On Wed 28th October world leaders from 29 countries attended a virtual meeting hosted by Finnish President and Norwegian Prime Minister. This followed up on the “Call for Control of Frontier AI Models” from last month, calling for a global AI oversight body.10,11

Political leaders from 29 countries across the world discussed international cooperation in AI risk management, and their vision the path to establishing common standards.10

The UK has not signed and did not attend the meeting.11

Countries backing the Call
  1. 2221 Seplaunch
  2. 3123 Octlatest
  3. 28 Oct meeting29 countries

UK: NOT SIGNED

10,11

Good news!

400 more patients a month learn of a suspected lung cancer within 24 hours thanks to AI helping NHS clinicians read chest X-rays.2

11.5% of patients receive faster diagnosis and wait for follow-up CT scans have fallen.2

A pencil drawing of a chest X-ray, a spot on one lung flagged by AI

Next week key dates

    • Lords 3rd reading
    • Sovereign AI Summit
    • US midterms
    • NI consultation closes
  • 2 Nov: Lords third reading of the Cyber Security and Resilience Bill, and the UKAI Sovereign AI Summit.7,12
  • 3 Nov: US midterms.
  • 4 Nov: Northern Ireland’s AI strategy consultation closes.13

Further ahead · Fri 13 Nov Second reading of Alex Sobel MP’s Artificial Superintelligence Bill, a ten-minute rule bill to ban superintelligent AI.14

Sources

  1. Anthropic, PBC (2026) Form S-1 registration statement. Filed with the US Securities and Exchange Commission, 26 October. ‘Risk Factors’, pp. 28 and 46. Available at: https://www.sec.gov/Archives/edgar/data/0002130357/000162828026067168/anthropicpbc-sx1.htm (Accessed: 30 October 2026).
  2. HM Treasury, Jarvis, D., Healey, J. and Narayan, K. (2026) Chancellor unveils Britain’s ambitious plan for AI. GOV.UK, 24 October. Available at: https://www.gov.uk/government/news/chancellor-unveils-britains-ambitious-plan-for-ai (Accessed: 30 October 2026).
  3. City A.M. (2026) ‘Budget 2026: Healey bets £700m on AI to boost British tech firms’, City A.M., 28 October. Available at: https://www.cityam.com/budget-2026-healey-bets-700m-on-ai-to-boost-british-tech-firms/ (Accessed: 30 October 2026).
  4. HM Treasury and Healey, J. (2026) Budget 2026 speech. GOV.UK, 28 October. Available at: https://www.gov.uk/government/speeches/budget-2026-speech (Accessed: 30 October 2026).
  5. Cabinet Office, AI Security Institute, National Cyber Security Centre and DCMS (2026) Britain steps up national defences against growing AI risks. GOV.UK, 12 October. Available at: https://gov.uk/government/news/britain-steps-up-national-defences-against-growing-ai-risks (Accessed: 30 October 2026).
  6. Hansard (2026) Cyber Security and Resilience (Network and Information Systems) Bill, Report stage. HL Deb, 26 October. Available at: https://hansard.parliament.uk/Lords/2026-10-26/debates/B413EFAE-E7D4-4742-986D-092884E0D8D1/CyberSecurityAndResilience(NetworkAndInformationSystems)Bill (Accessed: 30 October 2026).
  7. Resultsense (2026) ‘Ministers kept AISI voluntary to protect its access. Peers voted for a review’, Resultsense Insights, 27 October. Available at: https://www.resultsense.com/insights/2026-10-27-aisi-voluntary-access-lords-review-cyber-bill/ (Accessed: 30 October 2026).
  8. Carmarthenshire News Online (2026) ‘Ben Lake MP to lead first Commons debate on regulating advanced AI’, 26 October. Available at: https://www.carmarthenshirenewsonline.com/nationalnews/ben-lake-mp-to-lead-first-commons-debate-on-regulating-advanced-ai/ (Accessed: 30 October 2026).
  9. NPR (2026) ‘Inside Anthropic’s IPO + How AI Agents Spy on Your Messages’, Machine Gods [podcast transcript], October. Available at: https://www.npr.org/transcripts/nx-s1-6019724 (Accessed: 30 October 2026).
  10. Government of Norway (2026) Joint statement by President of the Republic of Finland Alexander Stubb and Prime Minister of Norway Jonas Gahr Støre on how to follow up the call for control over development of frontier AI models. Regjeringen.no, October. Available at: https://www.regjeringen.no/no/whats-new/joint-statement-how-to-follow-up-the-call-for-control-over-development-of-frontier-ai-models/id3186489/ (Accessed: 30 October 2026).
  11. Office of the President of the Republic of Finland (2026) President Stubb and Prime Minister of Norway: the rapid development of AI requires international cooperation, 21 September (signatory list updated 23 October). Available at: https://www.presidentti.fi/en/president-stubb-and-prime-minister-of-norway-the-rapid-development-of-ai-requires-international-cooperation/ (Accessed: 30 October 2026).
  12. All AI Events (2026) Sovereign AI Summit, 2 November, London (UKAI events calendar). Available at: https://allai.events/event/sovereign-ai-summit (Accessed: 30 October 2026).
  13. The Executive Office (2026) Northern Ireland’s draft Artificial Intelligence Strategy: public consultation, 12 August – 4 November. Available at: https://www.executiveoffice-ni.gov.uk/consultations/northern-irelands-draft-artificial-intelligence-strategy (Accessed: 30 October 2026).
  14. Hansard (2026) Artificial Superintelligence: motion for leave to bring in a Bill (Alex Sobel MP). HC Deb, 8 September. Bill 142, second reading set for 13 November.