CCAO-F domain - 15% of the exam

Governance, Risk, and Responsible Use

Governance, Risk, and Responsible Use is 15% of the CCAO-F exam. These are the objectives it covers, each with practice questions, with every answer explained.

The domain in numbers

43
Practice questions
4
Objectives
9
Easy
34
Medium

Where people go wrong

  • Treating an instruction to Claude as a data control

    The wrong answers ask Claude to forget the data, ignore the details, or promise not to repeat them, and treat that as a safeguard. The data has already been entered into the tool, which is what the policy prohibits. An instruction changes what Claude writes back, not what was shared. The correct answers keep the data out first, by removing or replacing it before upload, and report an accidental upload to the data protection officer as the policy requires.

    See it in a sample question
  • Treating partial removal of identifiers as anonymisation

    The wrong answers delete names, or swap them for codes, and call the file anonymous. Identifying detail remains: a job title, depot and leaving month in a small team, a quote from the only wheelchair user on Elm Terrace, or a lookup sheet that reverses the codes. Coded data with its key is pseudonymised and is still personal data. The correct answers generalise indirect identifiers, review quoted text before publishing and leave the key out.

    See it in a sample question
  • Letting Claude make the decision about a person

    The wrong answers have Claude rank, score or approve people, then add a safeguard that still leaves the decision with Claude: written reasons, a panel adjusting its scores, a batch sign-off that never opens a case, or scores feeding straight into bonuses. The policy puts a named person or panel in charge of each decision. The correct answers use Claude to summarise or draft guidance while people decide, and the reviewer examines each case.

    See it in a sample question
  • Abandoning Claude where the policy allows it

    The wrong answers refuse or step back from Claude entirely: no AI for anything involving staff, no use of the tool on pupil notes, no use for confidential client material. Each policy permits the use under a condition, such as expert review, removing identifying details, or working in the firm's managed account. Abandoning the tool puts the deadline at risk and removes no real risk. The correct answers meet the condition and carry on.

    See it in a sample question

Objectives in this domain

What to study

The official documents this domain's practice questions cite most.

Sample questions from this domain

Free sampleGovernance, Risk, and Responsible Usemedium

A secondary school teacher has 90 end-of-year reports to write in two weeks. The school's AI-use policy allows staff to use Claude to draft text but prohibits entering pupils' names, special educational needs details or other identifying information. The teacher's notes on each pupil include all of these. What should the teacher do?

  • APaste the notes in full, but instruct Claude not to store or reuse any pupil information once the reports are finished
  • BRemove identifying details from the notes, use Claude to draft the wording, then add specifics and check each report by hand Correct
  • CWrite all 90 reports by hand, because any use of Claude on pupil reports breaches the school's AI-use policy
  • DPaste the notes with names replaced by initials, since initials alone are not enough to identify a pupil
An appropriate task can be done inappropriately; change how you use Claude so the input meets policy instead of abandoning the task. Report drafting suits Claude, but this policy makes the method matter as much as the task: identifying pupil information must not go in. Removing it before drafting, then adding specifics outside Claude, keeps the useful work while respecting the policy. Partial measures such as initials or a request not to retain data leave prohibited information in the input.

Why A is wrong: Asking Claude to forget the data feels like a safeguard. But the policy forbids entering the information at all, and an instruction to the model is not a control that undoes having shared it.

Why B is correct: Drafting report wording is a permitted use, and stripping identifying details first keeps the input within policy. Adding the specifics and checking each report afterwards keeps the teacher accountable for what parents read.

Why C is wrong: Avoiding the tool entirely feels like the safest reading of the policy. But the policy allows drafting, so giving up over-corrects when removing the identifying details would make the task permitted.

Why D is wrong: Swapping names for initials looks like anonymisation. But the special educational needs details are still included, which the policy prohibits, and initials combined with class details can still identify a pupil.

Free sampleGovernance, Risk, and Responsible Usemedium

An HR adviser at a regional logistics company wants Claude to identify common themes across 14 exit interview notes from the past year. The company's AI-use policy forbids entering information that could identify an individual employee. Each note names the leaver, and most also give the person's job title, depot and leaving month. Several depots have only one or two staff in a given role. What should the adviser do before uploading the notes?

  • ARemove the leavers' names and upload the rest, because without names the notes no longer identify anyone
  • BUpload the notes unchanged and instruct Claude not to mention any names or roles in its theme summary
  • CRemove the names and generalise job titles, depots and leaving dates that could point to a specific person Correct
  • DAsk Claude to anonymise the notes in a first chat, then paste its anonymised version into a fresh chat
Anonymisation must remove indirect identifiers, such as role, location and date combinations, as well as names when a small group makes individuals recognisable. A person can be identified by a combination of details even when their name is gone, and the risk is highest in small populations where one role at one site in one month describes a single individual. Effective anonymisation therefore removes direct identifiers and generalises or drops indirect ones before the data enters the tool, because what happens after upload cannot undo the disclosure.

Why A is wrong: Names are the obvious identifier, so stopping there feels sufficient. It is wrong here because the stem says some depots have only one or two people in a role, so title, depot and month together still point to a specific leaver.

Why B is wrong: It is tempting because the summary would look clean. But the identifying data has still been entered into the tool, which is exactly what the policy forbids; an instruction to the model is not a data control.

Why C is correct: The policy covers any information that could identify an employee, and in small teams a combination of role, location and date does that on its own. Removing direct identifiers and generalising the indirect ones keeps the themes usable while meeting the policy.

Why D is wrong: Using Claude to do the anonymising feels efficient. It fails because the original identifying notes are uploaded in the first chat, so the policy is breached before any anonymisation happens.

Free sampleGovernance, Risk, and Responsible Usemedium

An HR coordinator at a logistics company has 300 applications for warehouse supervisor roles and a shortlist due in three days. The company's recruitment policy states that every shortlisting decision must be made by a named recruiter. The hiring manager suggests asking Claude to rank the applicants and reject the bottom half automatically. What should the coordinator do?

  • AUse Claude to summarise each application against the published criteria, and have the recruiter make every shortlisting decision Correct
  • BLet Claude reject the bottom half, because the published criteria are objective and it applies them more consistently
  • CAsk Claude to rank the applicants and write a reason for each rejection, then send the rejections as drafted
  • DDrop Claude from the process entirely, since any use of AI in recruitment breaches the company's recruitment policy
Claude can prepare and summarise information for a hiring decision, but the decision about each candidate stays with an accountable person. Shortlisting is a decision about a person, so it needs a human who can be held accountable and who can spot when the output misreads someone. Summarising applications against stated criteria is drafting and analysis work that Claude does well, and it speeds up the recruiter without transferring the decision. The policy fixes who decides, which separates an appropriate supporting use from an inappropriate automated one.

Why A is correct: This uses Claude for what it suits, condensing 300 applications against the criteria so the recruiter can work faster, while the shortlisting decision about each person stays with the named recruiter as the policy requires.

Why B is wrong: Consistency sounds like fairness, and the deadline makes automation tempting. But a rejection is a decision about a person, the policy requires a named recruiter to make it, and Claude can misread or unevenly weigh applications in ways nobody would catch if no human looks.

Why C is wrong: Written reasons look like accountability, so this can feel more responsible than a bare ranking. Claude is still making the decision, the reasons are generated after the fact rather than checked, and no recruiter has decided anything as the policy requires.

Why D is wrong: Caution about AI in hiring is sensible, so stepping back can feel safe. But the policy governs who makes the decision, not whether a recruiter may use help to read applications, so abandoning Claude over-corrects and puts the deadline at risk for no gain.

Other domains in this exam

See also the CCAO-F cert hub, the study guide, and the cheat sheet.

Examworthy is not affiliated with or endorsed by Anthropic. Original, blueprint-aligned practice material only.