DBS publishes algorithmic transparency record for Barring Referral AI Summary Tool
AI-generated. This page was written and translated automatically by AI and checked by software against the cited sources. No human editor reviewed it. Not legal advice.
Summary
The Disclosure and Barring Service has published an algorithmic transparency record describing an AI tool that generates summaries of barring referrals for caseworkers. The tool uses foundation models hosted on AWS Bedrock, with citations linked to source evidence, to help triage caseworkers navigate referral documents. The record states the tool makes no automated decisions and caseworkers must verify summaries against the original evidence. Reply, the existing technical provider, develops and maintains it. The record says the tool will be used by a triage team of about 16 people handling about 1,200 cases per month. It describes testing in Alpha and Private Beta phases, a Data Protection Impact Assessment, an Equality Impact Assessment that found no disproportionate adverse impact in Private Beta testing, and listed risks with mitigations such as bias, hallucination and over-reliance.
Why it matters
The record shows how a UK public body discloses its use of generative AI in safeguarding-related casework. It states that human caseworkers retain all decisions, that the existing DBS appeals process applies, and that the data protection assessment will be reviewed every 12 months. Caseworkers, referred individuals and the supplier are affected.
- Parties involved
- Disclosure and Barring Service; Cabinet Office; Department for Science, Innovation and Technology; Government Digital Service; Reply
- Primary source
- GOV.UK (gov.uk)
Source retrieved: . Sources are often in other languages; links open the original. - Topics
- Government use of AI · Transparency & labeling · Privacy & automated decisions