Concept 3 of 5

AI governance and compliance

3 questions test this

Governance is the arrangement that makes an AI system accountable to somebody after it is running. The syllabus splits it into the standards you are measured against, the data practices underneath, and the routine that keeps both alive.

The standards

ISO publishes the management system standards. ISO/IEC 42001 is the one for artificial intelligence, describing a management system for developing and using AI responsibly, and it sits alongside ISO/IEC 27001 for information security. Certification against it is how an organisation demonstrates a governance system exists rather than asserting it.

SOC reports, System and Organization Controls, are audits of a service organisation's controls. SOC 2 is the familiar one for security, availability, processing integrity, confidentiality and privacy, and it is what a customer's procurement team asks for.

Algorithm accountability laws are the growing category of regulation requiring disclosure, impact assessment or a right to explanation where automated decisions affect people. The specifics vary by jurisdiction and the direction is consistent, which is that a system deciding something about a person is expected to be documented, testable and appealable.

Neither exam tests the text of any regulation. Both test whether you know which kind of instrument a given name is.

Data governance

Data lifecycle covers a record from collection to deletion, and the useful question is who authorised each stage. Data collected for one purpose and used to train a model is a different purpose, and consent frequently does not travel.

Retention is how long you keep it. Longer retention is a larger breach and a larger obligation, and training data has a habit of outliving the reason it was gathered.

Residency is where it physically lives. For AI this reaches further than it looks, because inference has to happen in the permitted region too, and not every model is available in every region.

Logging, monitoring and observation are what make an incident reconstructable. For a generative system that means recording prompts, retrieved context and responses, which immediately becomes its own privacy problem, since the log now contains whatever the user typed.

The governance routine

Policies state what the organisation permits, and the useful ones are specific. Which categories of decision may be automated, which require a person, and what data may reach a third party model.

Review cadence is how often a deployed system is looked at again. Bias drifts, data shifts, and providers update models underneath you, so a system signed off once is not governed.

Review strategies cover what a review consists of. Performance by subgroup, sampled human evaluation, incidents since the last review, and whether the documented limitations still match reality.

Transparency standards decide what gets published. Model cards, disclosure that a user is talking to a system, and stated limitations.

Team training appears in the syllabus and is easy to overlook. People deploying and operating these systems need to understand the failure modes, and an untrained team is the most reliable governance gap there is.

The Generative AI Security Scoping Matrix

Named explicitly in the AWS exam guide. It is an AWS framework that places a generative AI use case in one of five scopes, running from consumer applications you merely use, through building on a hosted model, to training your own from scratch. Security obligations differ sharply along that line, and the matrix exists so a team can work out which ones are theirs.

The four kinds of tool, whichever platform you are on

Governance tooling divides into four jobs, and every platform has a product for each. Holding the job rather than the product name is what transfers.

Configuration compliance, evaluating your own resources against rules. AWS Config, Google Security Command Center posture management, Azure Policy.

Evidence collection, gathering proof continuously against a framework so an audit is not a scramble. AWS Audit Manager, Azure Compliance Manager, and Google Assured Workloads for regulated environments.

The provider's own compliance posture, meaning their SOC and ISO reports rather than yours. AWS Artifact, Google Compliance Reports Manager, Azure Service Trust Portal.

The activity log, recording who called what and when, which is what makes an incident reconstructable. AWS CloudTrail, Google Cloud Audit Logs, Azure Monitor activity logs.

The pattern worth carrying anywhere is that the first two concern your configuration, the third concerns the provider's, and the fourth is the record of what actually happened.

Practise this

You need one model or feature that is serving real traffic, a stopwatch, and permission to interrupt colleagues.

Start the clock and answer five questions, writing the elapsed minutes beside each as the answer arrives. Count the minutes rather than the effort, and mark where each answer came from honestly, because something a colleague remembered is a different artefact from something you were able to read.

Model or feature .................................
Clock started at ......

  Q1  Which exact version is serving traffic now?      ___ min
  Q2  Who approved it going live, by name?             ___ min
  Q3  What was it evaluated against before that
      approval?                                        ___ min
  Q4  On what date, and has it been rerun since?       ___ min
  Q5  Where is each of the answers above written
      down?                                            ___ min

  Mark every answer
    W  written down somewhere I can link to
    R  somebody remembered it
    X  could not answer at all

Total elapsed ______ minutes

The elapsed time is the finding rather than the answers. Governance that works produces all five within a couple of minutes from one place, and the usual result is a scavenge across a deployment tool, a chat thread, a spreadsheet and somebody's memory, which is the same search an auditor, a customer's security questionnaire or an incident at nine in the evening would demand at a far worse moment. Every answer marked R is one resignation away from becoming an answer marked X.

The exercise teaches you that governance is measured by how quickly the record answers, rather than by whether a policy document exists.

Common misconceptions

Using a compliant cloud provider makes your AI system compliant.

The provider's certifications cover the provider's infrastructure. What you built on it is assessed separately, and inheriting a control is something you have to document rather than assume.

Governance is paperwork that slows delivery down.

The parts that matter are decisions you would have to make anyway. What the system is allowed to decide, who owns it, what triggers a review, and when it gets switched off. Teams that skip them make the same decisions later under pressure.

Data residency is satisfied by storing data in the right region.

Inference has to happen there too. A request routed to a model hosted elsewhere moves the data, which is why regional availability of a particular model is a compliance question rather than a convenience one.

3 questions test this concept

A company must keep customer data within a jurisdiction. Data is stored in a compliant region and a generative feature calls a model hosted elsewhere. What is the position?

  • AResidency is satisfied, since the system of record stays in the permitted region.
  • BResidency is satisfied provided the response is not stored outside the region.
  • CResidency is breached, because inference sends the data outside the permitted region.
  • DResidency is satisfied because inference is transient rather than storage.
Check whether it stuck.

One per page, with a worked explanation.

Start the set
Related material
Book
Data and Goliath, On retention, residency and the case for collecting less.
Template
Launch checklist, Everything that has to happen from two weeks out to one week after release, grouped by when it falls due, each line with a named owner and a go or no go decision on the day.
Template
Stakeholder map, One row per stakeholder with what they need from the product, their influence and interest scored one to five, where they stand today, where you need them to be, and who owns the next conversation.