PiR2-IT
Start an engagement
PiR2-IT
Start an engagement
Ethics and fairness services for AI used in banking, public services and industry — from bias testing and explainability to ethics impact assessments and the evidence regulators ask for. 36 implementations to date.
Postcode, age, gender or employment history can act as proxies and produce unfair outcomes even when protected attributes are removed.
Credit scoring, access to public benefits and recruitment are high-risk uses under the EU AI Act; regulators expect documented testing, oversight and explanations.
Principles are adopted, but no one measures fairness after go-live, and model drift quietly reintroduces bias.
Measurement of disparate impact, equal opportunity and calibration across groups, root-cause analysis of proxies, and mitigation before and after deployment.
Explanations that a customer, a citizen or a case officer can understand, with a route to challenge an automated decision and obtain human review.
Fundamental-rights and ethics impact assessments for high-risk AI, covering affected groups, harms, safeguards and human oversight.
Ethics review boards, approval gates, fairness KPIs in production, drift alerts and audit-ready evidence for supervisors.
Credit scoring, loan origination, fraud detection, AML alerts, pricing and collections — aligned with the EU AI Act, EBA guidelines and consumer-protection rules.
Eligibility for benefits, tax and inspection risk scoring, case prioritisation and citizen services, where decisions must be fair, explainable and contestable.
Recruitment and workforce analytics, insurance underwriting, quality control and customer service automation.
Ethics and fairness work is grounded in research, international standards and public policy, not in a single vendor’s view.
Joint work on fairness metrics, bias detection, explainability methods and testing on real-world data, so methods are evidence-based and peer-reviewable.
Participation in ISO, CEN and ETSI standardisation work and in IEEE and ITU working groups on AI and cybersecurity — including the ISO/IEC 42001 and IEEE 7000 families of AI standards.
Collaborating member of the OECD Working Party on AI Governance (AIGO) and contributor to the OECD.AI Policy Observatory, involved in the OECD AI Principles since their drafting began in 2017.
Practical governance models — ethics boards, review gates, human oversight and audit evidence — that turn principles and standards into controls institutions can operate.
Current state, delivery friction, architecture debt and the decisions that must be made first.
Reference architecture, governance model, control points, sequencing and evidence structures.
Design authority, prototype validation and assurance reviews as the programme proceeds.
A structured test of whether an AI system produces systematically different outcomes for different groups, why it happens, and how to correct it — documented so it can be shown to a regulator.
For high-risk AI under the EU AI Act — including credit scoring and access to essential public services — risk management, data governance, transparency and human oversight are mandatory. Ethics and fairness work produces that evidence.
Yes. Independent second-opinion reviews of third-party AI are a large part of this work.
36 to date, across banking, the public sector and industry. References are available on request, under NDA.
Yes. Ethics is built into every module of the engine, and the same fairness controls are applied to client systems.
PiR2-ITIndependent consulting for institutions, regulated sectors and high-trust programmes.