GBR-003

DWP Generative-AI Pilots

Download PDF
United Kingdom Europe & Central Asia High income Pilot / Controlled Trial Phase Confirmed

Department for Work and Pensions (DWP) Digital

At a Glance

What it does LLMs for content creation, transformation and modality conversion — Operational and process automation
Who runs it Department for Work and Pensions (DWP) Digital
Programme DWP Generative-AI Pilots
Confidence Confirmed
Deployment Status Pilot / Controlled Trial Phase
Key Risks Not assessed
Key Outcomes Reported reduction in documentation time and improved internal communication efficiency during the 2024–2025 pilot period.
Source Quality 3 sources — Government website / press release, News article / media

The Department for Work and Pensions (DWP) in the United Kingdom has been conducting a series of generative AI pilot projects through a structured 'lighthouse' programme established in 2023. The programme was designed 'to test and learn in a safe and governed environment where all types of AI can be used to assist us in the delivery of our customer outcomes and department efficiencies,' according to the DWP's contract with IBM.

On 1 August 2024, the DWP entered into an initial five-month contract with IBM to support the department in delivering 'gen AI lighthouse projects.' The contract can be extended for five further months, up to the end of May 2025, and will be worth £10.8 million inclusive of VAT if it runs for its full potential term. IBM is providing the DWP with 'application development and integration' services, with the work to be set out in 'a series of statements of works' covering 'strategy and control,' 'lighthouse' projects, and a minimum viable product for what appears to be a Generative AI Lab facility.

The DWP planned to launch five generative AI projects with director general (DG) sponsorship during the 2024-25 financial year. As stated in the IBM contract: 'We are scaling up our work in generative AI during this financial year. We have identified a number of candidate areas and will launch five projects with DG sponsorship during the current financial year with the key test that they have a high likelihood of being scalable into the overall business.' The agreement did not specify the details of these five projects.

The department's annual report for 2023-24 claimed that during the year, the DWP had 'rapidly and successfully tested multiple generative AI proofs of concept.' Three specific tools were highlighted. 'Whitemail' is a technology that 'scans documents and quickly identifies vulnerable customers, allowing DWP to fast track and intervene.' 'AIgent' supports agents administering Personal Independence Payments 'by summarising evidence for inclusion in decision letters.' The 'A-cubed' system 'provides work coaches with quick access to advice to help to support customers move closer to the labour market.'

The then work and pensions secretary Mel Stride stated in late 2023 that the lighthouse scheme was 'exploring the use of AI in several use cases which include: trialling AI-enabled projects to complement the services work coaches provide to customers in job centres; trialling how AI can write, update, or organise code to address the current digital skills shortage in areas like software engineering; and trialling productivity tools for all colleagues to use, such as rapidly summarising policy documents or providing simple tools to gather information for frontline colleagues.'

The programme's governance structure involves three aspects according to the IBM contract: 'steering from across the department – [because] it's not just [about] digital; enablement – ensuring that we have the capabilities to make best use of the opportunity while guarding against its risks; [and] transformation – launching value-add digital projects.' The contract explicitly states that 'the programme will consider ethical, legal and commercial enablers and ensure that DWP continues with its overall strategy of keeping humans in the loop on any decision.'

The capability development is being conducted within the DWP's existing Innovation Lab. Despite the growing focus on generative AI, the DWP effectively banned its employees from using external tools such as ChatGPT in the course of their work or via government-issued devices. The department has instead opted to pursue the use of internal tools, largely based on Microsoft Copilot technology.

Classifications follow the DCI AI Hub Taxonomy. Hover over field labels for definitions.

Social Protection Functions

Implementation/delivery chain
Case management primaryProfiling, job matching and support services
SP Pillar (Primary) The social protection branch: social assistance, social insurance, or labour market programmes. Social assistance
SP Pillar (Secondary) The social protection branch: social assistance, social insurance, or labour market programmes. Social assistance
Programme Name DWP Generative-AI Pilots
Programme Type The type of social protection programme, classified under social assistance, social insurance, or labour market programmes. View in glossary Other
System Level Where in the social protection system the AI is applied: policy level, programme design, or implementation/delivery chain. View in glossary Implementation/delivery chain
Automation Subtype For operational automation cases: (a) document processing and generative staff assistance, or (b) workload and resource forecasting. (a) Document processing and generative staff assistance
Programme Description The DWP Generative AI Pilots programme is a series of AI experiments conducted through the department's 'lighthouse' programme and Innovation Lab, supported by a £10.8m contract with IBM. The programme tests generative AI tools for administrative productivity including document scanning, evidence summarisation for benefit decisions, work coach support, policy document summarisation, and code writing assistance. Three named proof-of-concept tools were developed during 2023-24: Whitemail (document scanning for vulnerable customers), AIgent (evidence summarisation for PIP decisions), and A-cubed (work coach advice assistance). The programme uses Microsoft Copilot-based internal tools rather than external services like ChatGPT.
Implementation Type How the AI output is produced: Classical ML, Deep learning, Foundation model, or Hybrid. Affects validation, compute requirements, and governance profile. View in glossary Foundation model
Lifecycle Stage Current stage in the AI lifecycle, from problem identification through to monitoring, maintenance and decommissioning. View in glossary Integration and Deployment
Model Provenance Origin of the AI model: developed in-house, adapted from open-source, commercial/proprietary, or accessed via third-party API. View in glossary Commercial/proprietary
Compute Environment Where the AI system runs: on-premise, government cloud, commercial cloud, or edge/device. View in glossary Not documented
Sovereignty Quadrant Classification of data and compute sovereignty: I (Sovereign), II (Federated/Hybrid), III (Cloud with safeguards), or IV (Shared Innovation Zone). View in glossary Not assessed
Data Residency Where the data used by the AI system is stored: domestic, regional, or international. View in glossary Not documented
Cross-Border Transfer Whether data crosses national borders, and if so, whether documented safeguards are in place. View in glossary Not documented
Decision Criticality The rights impact of the decision the AI supports. High criticality requires HITL oversight; moderate requires HOTL; low may operate HOOTL. View in glossary Moderate
Human Oversight Type Level of human involvement: Human-in-the-Loop (active review), Human-on-the-Loop (monitoring), or Human-out-of-the-Loop (periodic audit). View in glossary HITL
Development Process Whether the AI system was developed fully in-house, through a mix of in-house and third-party, or fully by an external provider. View in glossary Mix of in-house and third-party
Highest Risk Category The most significant structural risk source identified: data, model, operational, governance, or market/sovereignty risks. View in glossary Not assessed
Risk Assessment Status Whether a formal risk assessment, informal assessment, or independent audit has been conducted for this system. Not assessed

Risk Dimensions

Governance and institutional oversight risks
Market, sovereignty and industry structure risks
Operational and system integration risks

Impact Dimensions

Autonomy, human dignity and due process
  • Human oversight protocol
CategorySensitivityCross-System LinkageAvailabilityKey Constraints
Beneficiary registries and MISPersonalSingle source (no linkage)Currently available and usedCase management records used as input for AI summarisation and drafting tools; contains personal claimant details including benefit history and needs assessments
Unstructured and text-based contentPersonalSingle source (no linkage)Currently available and usedInternal administrative text including case summaries, correspondence drafts, and policy notes; access restricted to DWP staff within government-secured cloud environment

GOV.UK (2025) 'Landmark government trial shows AI could save civil servants nearly 2 weeks a year', Department for Science, Innovation and Technology, 2 June. Available at: https://www.gov.uk/government/news/landmark-government-trial-shows-ai-could-save-civil-servants-nearly-2-weeks-a-year (Accessed: 22 March 2026).

View source Government website / press release

Trendall, S. (2024) 'EXCL: DWP and £11m supplier prep five gen AI projects "with director general sponsorship"', PublicTechnology, 4 December. Available at: https://www.publictechnology.net/2024/12/04/society-and-welfare/excl-dwp-and-11m-supplier-prep-five-gen-ai-projects-with-director-general-sponsorship/ (Accessed: 22 March 2026).

View source News article / media

Trendall, S. (2025) 'DWP seeks leader to build gen AI team and plots £12m deal to target "to be identified business problems"', PublicTechnology, 27 February. Available at: https://www.publictechnology.net/2025/02/27/government-and-politics/dwp-seeks-leader-to-build-gen-ai-team-and-plots-12m-deal-to-target-to-be-identified-business-problems/ (Accessed: 22 March 2026).

View source News article / media
Deployment Status How far the system has progressed into real-world operational use, from concept/exploration through to scaled and institutionalised. View in glossary Pilot / Controlled Trial Phase
Year Initiated The year the AI system was first initiated or development began. 2024
Scale / Coverage The scale and geographic or population coverage of the deployment. Pilot-scale within DWP; five projects planned for 2024-25 financial year with scalability as key criterion
Funding Source The source(s) of funding for the AI system development and deployment. Government (DWP budget via IBM contract worth £10.8m inclusive of VAT)
Technical Partners External technology vendors, academic partners, or development partners involved. IBM (£10.8m contract for gen AI lighthouse projects, application development and integration); Microsoft (Copilot technology for internal productivity tools); DWP Innovation Lab (internal development environment).
Outcomes / Results Reported reduction in documentation time and improved internal communication efficiency during the 2024–2025 pilot period.

How to Cite

DCI AI Hub (2026). 'DWP Generative-AI Pilots', AI Hub AI Tracker, case GBR-003. Digital Convergence Initiative. Available at: https://socialprotectionai.org/use-case/GBR-003 [Accessed: 1 April 2026].

Change History

Created 30 Mar 2026, 08:39
by v2-import (import)