Credit decisions, hiring algorithms and the right to an explanation
Four stories about AI decisions that affect people's lives, and what the law says about proving they were fair.
- 01
Lenders must explain AI credit decisions
New EU rules strengthen consumer rights when AI is used for credit decisions. Lenders must be able to explain and review decisions made by AI systems.
Why it matters for evidence: If someone challenges a decision, the lender needs to show what the AI did and why, from logs that hold up.
Source: finanstid.se (in Swedish)
- 02
US court allows AI hiring discrimination case
A federal judge has allowed a class-action lawsuit against Workday to proceed, alleging its AI recruitment tools discriminate against applicants over 40. The case centres on whether the company can show its systems operated fairly.
Why it matters for evidence: To defend an AI hiring decision in court, the company needs a record of what the system did and how.
Source: facebook.com
- 03
Three pillars for AI governance
AI governance requires a named owner for each AI decision, review proportionate to its risk, and a record of what was accepted. This framework enables boards, auditors and regulators to verify accountability.
Why it matters for evidence: The third pillar, a record of what was accepted, only works if it can be shown unchanged later.
Source: dataproducts.io
- 04
Machinery Regulation brings logging duties
From 20 January 2027, the EU Machinery Regulation requires safety-relevant software changes to be logged and traceable for five years, with protection against corruption.
Source: facebook.com
- 05
Public agencies must prove AI is safe
Public agencies must prove AI systems are safe, fair and accountable before purchase and ongoing use, through documented assurance processes including risk management, logging and impact assessments.
Source: publicsectornetwork.com
- 06
AI scribes in UK clinics need compliance
AI scribe tools in UK clinical practice are classified as medical devices by the MHRA and must comply with GDPR and data residency requirements.
Source: tandemhealth.ai