Anthropic wins UK deal to pilot Claude-powered GOV.UK assistant for job seekers

The UK has tapped Anthropic to pilot a Claude-powered assistant on GOV.UK, starting with employment support. It will guide multi-step tasks, keep context, and scale if it works.

Categorized in: AI News Government
Published on: Jan 29, 2026
Anthropic wins UK deal to pilot Claude-powered GOV.UK assistant for job seekers

Anthropic to build Claude-powered AI assistant for GOV.UK

The UK government has selected Anthropic to build and pilot an AI assistant for GOV.UK, starting with employment services. The tool will sit alongside existing services and run on Claude. The goal: help job seekers and people returning to work find roles, training options, and the right government support with less friction.

This work follows a 2025 Memorandum of Understanding between Anthropic and the UK government and will be delivered with the Department for Science, Innovation and Technology (DSIT). The rollout will follow the government's "Scan, Pilot, Scale" model to test, learn, and expand only when it's proven.

What it means for departments

Think of this as guided support on GOV.UK that can handle multi-step tasks, not just single questions. It can prompt users for details, adapt responses to their circumstances, and route them to the right services. For employment, that means better signposting across benefits, training, and job search support.

How the assistant will work

  • Built on Claude as an "agentic" system that can guide users through processes rather than answer one-off questions.
  • Adapts responses to individual situations and routes users to the right services based on their inputs.
  • Maintains context across interactions so returning users don't need to repeat themselves.
  • Gives users control over what the system remembers, with the option to opt out.
  • Handles personal data in line with UK data protection law.
  • Follows DSIT's phased approach: scan opportunities, pilot in the open, and scale if it works.

Safety and governance focus

Anthropic frames this as a safety-led deployment in partnership with DSIT. The project supports the government's AI Opportunities Action Plan and will use structured testing and evaluation before any wider rollout.

Anthropic says it continues to work with the UK AI Safety Institute on testing and evaluation to inform safeguards for public sector use.

Skills transfer and ownership

Anthropic will place engineers alongside civil servants and developers at the Government Digital Service. A key aim is to build AI and AI safety capability inside government so teams can run and maintain the system independently over time.

DSIT also sees this pilot as a way to gather evidence that improves governance and procurement approaches for future services.

First pilot: employment services

The initial live area focuses on job seekers and people returning to work. The assistant will give career guidance, highlight training options, and point people to the right services based on their situation. DSIT will oversee the build and pilot stages and decide on further expansion based on results.

UK footprint and current partnerships

Anthropic highlights ongoing collaboration with the UK AI Safety Institute and a growing London office across research, applied AI, policy, and commercial roles. The company cites work across British businesses-incident.io, Wordsmith, WPP, and London Stock Exchange Group-plus public sector and education partnerships with the London School of Economics, Iceland's Ministry of Education and Children, and the Rwandan Government.

What government teams can do now

  • Map common employment-related queries and tasks that cause drop-off or repeat contact.
  • Define success metrics: resolution rate, referral accuracy, time to complete, and user satisfaction.
  • Prepare content for structured guidance: clear steps, eligibility rules, and policy edge cases.
  • Line up privacy work: DPIA updates, data retention settings, opt-out flow, and user-facing notices.
  • Plan assurance: red-teaming, bias testing, accessibility checks, and safe fallbacks to human support.
  • Agree operational guardrails: escalation rules, audit logging, and clear ownership between DSIT, GDS, and service teams.
  • Upskill frontline and product teams on prompt practices, reviewing AI outputs, and handling exceptions.

Timeline and next steps

The assistant now moves into build and pilot under DSIT oversight, starting with employment services. If the pilot meets safety, quality, and value thresholds, it may expand to other areas on GOV.UK. Departments interested in future pilots should coordinate with DSIT and GDS to align use cases and readiness.

Optional upskilling for teams working with Claude

If your team is preparing to work with Claude in public services, a focused certification can speed up adoption while reducing risk.

Bottom line: this pilot is a practical step to improve how people find and use employment support on GOV.UK-safer, clearer, and easier to maintain at scale if it proves its value.


Get Daily AI News

Your membership also unlocks:

700+ AI Courses
700+ Certifications
Personalized AI Learning Plan
6500+ AI Tools (no Ads)
Daily AI News by job industry (no Ads)
Advertisement
Stream Watch Guide