1. Purpose
This policy sets out Murzo Group's approach to automated decision-making, profiling, AI-assisted decisions, scoring, ranking, recommendations, eligibility decisions, risk flags, and human review.
The purpose is to protect fairness, transparency, lawful processing, accountability, contestability, redress, and human oversight where AI or automation may affect people, customers, workers, suppliers, contractors, applicants, platform users, or third parties.
2. Scope
This policy applies to automated or AI-assisted decisions used in recruitment, right to work, vetting, worker monitoring, access control, customer support, complaints, pricing, credit or payment risk, fraud prevention, supplier approval, security operations, platform access, product recommendations, legal triage, and any decision that may materially affect a person or business.
It covers fully automated decisions, partly automated decisions, profiling, human-in-the-loop decisions, human-on-the-loop monitoring, algorithmic ranking, AI recommendations, and rules-based automation.
3. Core Principles
Automation must be proportionate, explainable enough for the context, fair, secure, lawful, accurate enough for the purpose, and subject to human accountability where needed.
- Automated systems must not be used to hide responsibility for a decision
- People should be told about automated processing where law or fairness requires it
- Significant decisions should have suitable safeguards, including human intervention where required
- Systems must not unlawfully discriminate or create unfair outcomes through biased data, proxy variables, poor design, or inappropriate use
4. Prohibited and Restricted Decisions
Murzo Group must not use solely automated decisions for employment dismissal, disciplinary action, legal claims, significant customer redress refusal, safety-critical action, biometric identification, sensitive vetting, or other legally significant decisions unless lawful, approved, and supported by appropriate safeguards.
AI outputs must not be treated as final evidence of dishonesty, misconduct, immigration status, criminality, safety risk, creditworthiness, mental state, political opinion, health, emotion, or intent without suitable human review.
5. Transparency and Notice
Where automated decision-making or profiling is used, Murzo Group should provide clear information about the purpose, logic at an appropriate level, data used, likely impact, human review route, and complaint route where required by law or fairness.
Notices must not overstate the accuracy, neutrality, independence, or authority of AI systems.
6. Human Review and Contestability
People affected by significant automated or AI-assisted decisions should have a practical route to raise concerns, request human review, provide additional information, correct inaccurate data, make representations, or complain where required by law or Murzo Group procedure.
Human reviewers must have enough authority, information, competence, and independence to challenge or change an automated outcome where appropriate. A human must not merely rubber-stamp an AI output.
7. Fairness, Bias and Accuracy
Automated systems should be assessed for accuracy, bias, discriminatory impact, inappropriate proxies, poor data quality, outdated assumptions, and unsuitable use in the relevant context.
Where errors or unfair outcomes are identified, Murzo Group may suspend, restrict, amend, retrain, replace, or withdraw the system and consider corrective action where appropriate.
8. High-Risk AI and International Requirements
Where automated decision-making may fall within EU high-risk AI categories, UK regulator expectations, Chinese AI or algorithm rules, or other local requirements, Murzo Group must consider role allocation, provider and deployer duties, human oversight, technical documentation, instructions for use, logging, transparency, data quality, monitoring, and serious incident escalation.
No person may deploy a high-impact automated decision system in Murzo Group's name without written approval.
9. Third-Party Systems
Suppliers of recruitment tools, fraud tools, scoring systems, chatbots, biometrics, analytics, security tools, customer platforms, and recommendation engines must be assessed for data use, fairness, transparency, auditability, security, support, complaint handling, and liability allocation where relevant.
Murzo Group does not accept responsibility for automated decisions made by third-party platforms, marketplaces, suppliers, employers, recruiters, finance providers, security providers, or affiliates outside Murzo Group's written approval or control, except where law requires otherwise.
10. Review and Evidence
Murzo Group may keep proportionate evidence of decision logic, approvals, notices, review requests, outcomes, supplier assurances, and complaints where needed for legal, privacy, HR, customer, security, or dispute purposes.
This policy should be reviewed when law, regulator guidance, automated systems, AI suppliers, decision processes, or Murzo Group activities change.
