4. Deployment and Governance

Governance, Fairness and Privacy

Translate accountability, fairness, privacy and current standards into system controls

Governance, Fairness and Privacy

Governance begins before modelling

“The data science team owns the model” is incomplete. A defensible system names:

  • a decision owner accountable for the policy;
  • a data owner accountable for meaning, access and retention;
  • a technical owner accountable for implementation and monitoring;
  • an independent reviewer able to challenge evidence and controls;
  • an affected person with a usable explanation and route to correction where appropriate.

Responsibility cannot be outsourced to a metric or vendor.

Use frameworks as questions

The voluntary NIST AI Risk Management Framework organises work into four functions:

FunctionHarborMart questionEvidence
governWho can approve, change or stop the delivery policy?roles, inventory, review cadence
mapWhich people, contexts and harms can the policy affect?use-case boundary and impact map
measureHow will validity, fairness, privacy and safety be tested?test protocol and results by group
manageWhich risks are accepted, mitigated, transferred or avoided?controls, owner, deadline and residual risk

NIST states that AI RMF 1.0 is being revised, so cite the version and review date. ISO/IEC 42001:2023 takes a management-system approach: policies, objectives, processes and continual improvement across an organisation. Neither framework turns a harmful objective into a good one; both require substantive judgement.

Fairness needs a decision and a harm

Suppose the system flags customers for manual delivery-window review. On a test set of 200 customers in each group:

GroupTPFNFPTNTPRFPRPPV
A3641614490.0%10.0%69.2%
B2713815267.5%5.0%77.1%

Both groups have about 90% accuracy, yet Group B experiences more missed late deliveries. Group A experiences more unnecessary reviews. No single column defines fairness because the harms differ.

Audit in this order:

  1. meaning: are groups and outcomes measured comparably?
  2. representation: who is absent, rare or poorly labelled?
  3. performance: calibration and relevant errors with uncertainty;
  4. allocation: who receives benefits, burdens and human review?
  5. recourse: can an affected person correct data or challenge an action?
  6. repair: change data, objective, workflow, threshold or product boundary—and retest consequences.

Small groups produce unstable rates. Report counts and intervals, not only percentages. A group-specific threshold may create new legal or ethical problems; obtain domain and legal review rather than treating it as an automatic technical fix.

Privacy by necessity

For each field, ask “what decision-relevant purpose requires this value?”

Proposed fieldTempting rationaleBetter control
exact date of birth“might improve prediction”use a justified age band or omit it
full postcodecaptures distancederive coarse distance, then restrict or delete source
free-text service notesrich contextexclude by default; may contain sensitive or unsafe text
customer identifierjoins outcomesreplace with controlled pseudonymous key

Minimisation reduces breach impact and spurious proxy use. It does not by itself make processing lawful. Specify purpose, lawful basis where relevant, access, retention, deletion, provenance and permitted downstream use with appropriate experts.

Security is part of analytic validity

Threats include altered source data, stolen credentials, poisoned labels, model extraction, prompt or input injection in AI-assisted workflows, and unauthorised policy changes. Controls should cover least privilege, secrets management, integrity checks, dependency review, immutable logs and tested incident response.

A perfectly calibrated model fed manipulated capacity data is an unsafe decision system.

Human review must be designed

“A human is in the loop” is not a safeguard unless the person has:

  • time and information to disagree;
  • authority to change the action;
  • guidance for difficult cases;
  • protection from automation bias and rubber-stamping incentives;
  • an override log reviewed for systematic patterns.

For a delivery disruption, show the reason, material evidence, alternatives and uncertainty—not an unexplained risk score.

The original EU AI Act, Regulation (EU) 2024/1689, was amended by Regulation (EU) 2026/1744. The European Commission's 31 July 2026 update reports that the amendment entered into force on 27 July 2026 and changed important high-risk-system timelines.

Therefore, never label a system or deadline from memory. Identify jurisdiction, role, intended purpose, affected people and the provision in force on the review date. This course offers analytical governance, not legal advice.

Quick check

HarborMart removes the customer-group field from training. Has it solved unfairness?

Answer
No. Other fields may act as proxies, historical labels may encode unequal service, and the policy may allocate burdens unevenly. Keep protected or audit attributes in a tightly controlled evaluation process where lawful and appropriate; removing them from model inputs does not remove the need to measure impact.

Next: Adoption and Business Value

Copyright © 2026