AGI in Public Services
How AI is being deployed in benefits, tax, immigration, healthcare, and other public services — and the trust, audit, and procurement requirements.
Executive summary
Governments are deploying AI in service delivery across benefits, tax, immigration, and healthcare. Outcomes are mixed: real productivity and access gains alongside high-profile failures. Procurement frameworks, audit requirements, and citizen trust are the binding constraints, not the underlying technology.
Key concepts
- Service delivery use cases
- Procurement and audit
- Citizen trust
- High-profile failures
- Capacity building
Where deployments work
Translation and accessibility, routine information services, internal productivity in government, and decision support for caseworkers under human oversight.
Where deployments fail
Automated decision-making without adequate human oversight, opaque models in high-stakes decisions, and procurement without audit have produced harmful failures (Robodebt and others).
What works
Procurement that requires evaluation and audit, human-in-the-loop on consequential decisions, clear redress paths, and ongoing capacity within government to oversee suppliers.
Public-sector capacity
Building genuine AI capacity inside government — not just procuring it — is now a strategic priority.
Key takeaways
- 01Service delivery gains are real and uneven.
- 02High-profile failures track unsupervised automation.
- 03Procurement and audit are the binding constraints.
- 04Internal capacity within government is essential.
Frequently asked questions
Is my government using AI on my data?
Increasingly likely. Transparency varies by jurisdiction; in the EU and UK there are growing disclosure requirements.
What can go wrong?
Wrongful denial of benefits, biased decisions, and lack of redress are the most documented harms. Oversight and audit are the protective layer.
Further reading
Related hubs
How AGI is reshaping policy, public services, democratic process, and geopolitics — and what citizens should expect.
The state of the global regulatory architecture for AI — what is in force, what is coming, and where the open debates are.
How AI-generated content, election integrity, and information environments are being affected — and the countermeasures being built.