Public Authority Algorithmic and Automated Decision-Making Systems Bill [HL]

Baroness Jones says the government is delivering transparency and public trust in public‑sector AI through the Algorithmic Transparency Recording Standard (ATRS), with 14 new records published and more to follow; ATRS will be mandatory for departments and ALBs for high‑impact, public‑facing tools. She also highlights GDPR safeguards (including Article 22) and reforms to the Data (Use and Access) Bill to enable responsible automated decision‑making, as the government plans to support safe and ethical AI adoption in public services.

The Committee highlights two main issues: first, in the Public Ownership Bill, a broad repealing power (to undo sections 30A/30B) could be misused and requires ministerial clarification on scope and procedure, with a preference for affirmative scrutiny unless tightly limited; second, in the Product Regulation and Metrology Bill, it condemns skeleton-style powers (clauses 1–3, 5–6 and 9) that would enable extensive, regulation-by-regulation controls—potentially including criminal-offence provisions—without adequate justification or scrutiny, and it calls for removing those delegations or providing a full rationale and stronger checks.

The Bill would require public authorities to assess and publicly report on algorithmic and automated decision‑making systems they develop or buy, including pre‑deployment Algorithmic Impact Assessments and post‑deployment Algorithmic Transparency Records. It introduces ongoing governance—independent scrutiny, mandatory bias assessment, logging, audits, staff training and human oversight—to promote fairness, privacy and due process, with an option to challenge or obtain redress. The Act applies to England and Wales (with some exclusions such as national security or purely calculative tools), comes into force six months after passage, and would be implemented by regulations laid by statutory instrument after consultation, alongside an independent dispute resolution service.