Jobs › Roles › AI Governance Analyst
What Is an AI Governance Analyst?
An AI governance analyst runs the day-to-day review of AI systems: keeping the AI inventory current, taking in new models and vendors, assessing their risk against the organization's policy, and preparing the evidence a decision-maker needs. The analyst asks what a system is for, who it affects and what evidence exists. The analyst does not build the system and does not approve it.
Updated November 4, 2026
What the role owns
The AI inventory, intake reviews for new models and vendors, and the risk assessments that go with them.
Who it reports to
An AI governance manager or the head of privacy.
A day in the job
Reviewing a new model or vendor against the policy, writing the risk assessment, chasing the owner for the documentation that should already exist, and keeping the AI inventory honest. Most weeks are reading and writing, not tooling.
What employers test for
Reading a model card and spotting what is missing. Writing clearly enough that a lawyer and an engineer both accept it. Knowing the NIST AI RMF and ISO/IEC 42001 well enough to cite the control, not the chapter.
How people get in
Privacy analysts and internal auditors move in fastest, because the job is evidence and judgment before it is technology. Engineers often find the writing harder than they expect.
What it gets confused with
Not a data scientist. The analyst asks what the model is for, who it affects and what evidence exists, and does not build it.
What the job produces
- Inventory records for each AI system: purpose, owner, data, vendor, risk tier and next review date.
- Intake reviews for new AI use cases and vendors.
- Risk assessments, and impact assessments where the policy requires them, with the supporting evidence attached.
- Decision packages for the committee or delegated approver that state the exact decision requested.
- A tracker of open conditions and re-review dates, each with an owner.
What the role can and cannot decide
The analyst recommends. On their own authority, the analyst does not approve a use, accept residual risk or stop a deployment. Those powers belong to a delegated approver or the AI governance committee, as the charter sets out. The analyst's real influence is the record: a decision cannot be made well, or audited later, without one.
How it differs from nearby roles
| Role | The question it answers | Typical output | Decision authority |
|---|---|---|---|
| AI governance analyst | Should this system proceed, and on what conditions? | Inventory, intake reviews, risk assessments, decision packages | Recommends. An approver or committee decides. |
| AI governance manager | Does the program exist, and does it work? | Policy, committee agenda, reporting upward | Owns the program and decides within delegated authority. |
| AI compliance analyst | Which rule applies, and can we prove we meet it? | Obligation register, evidence | Evidences controls. |
| AI risk or model risk analyst | Does the model perform as claimed, and how could it fail? | Validation, monitoring, findings | Challenges the model independently. |
| Data scientist or ML engineer | Can we build it, and does it work? | The model and its documentation | Builds and maintains the system. |
Titles vary between employers. Read the duties and the reporting line, not only the title.
A realistic example
A marketing team asks to use a vendor's generative AI tool to draft customer emails from purchase history. The analyst logs the request in the inventory, asks the vendor for its model documentation and data terms, checks with privacy whether customer data may leave the organization, and confirms whether drafts will reach customers without review. The assessment rates the use as moderate risk and recommends approval on three conditions: no sensitive personal data in prompts, human review before anything is sent, and a fresh review if the vendor changes the underlying model. The system owner approves under delegated authority, and the analyst records the decision, the conditions and the re-review trigger.
A portfolio project you can build
Fictional scenario, for practice only. Harbor Lane Credit Union, an invented 400-person financial cooperative, wants an AI tool that summarizes member complaints for its compliance team. Write the inventory entry, a two-page risk assessment that names the NIST AI RMF functions you used (Govern, Map, Measure, Manage), a decision package stating the exact decision requested, and a list of conditions with owners and re-review triggers. Mark every assumption you had to make.
Experience that transfers
Privacy analysts move in fastest because a privacy impact assessment and an AI risk assessment ask many of the same questions. Internal auditors bring evidence standards. Compliance staff bring policy mapping. Quality professionals bring document control and corrective action. Operations and program people bring what governance teams often lack: getting a busy owner to produce the document by Friday.
Where to go next
- How to become an AI governance analyst: the roadmap
- How to become an AI governance analyst: the long read
- Governance analyst: the entry point into AI governance
- AI governance analyst job description
- What is an AI inventory?
- What is an AI risk assessment?
- What is an AI vendor assessment?
- What is an AI governance committee?
- AI governance skills employers hire for
- Open AI governance jobs
AI Governance Analyst: tools, skills and workflows
What does an ai governance analyst do all day?
Reviewing a new model or vendor against the policy, writing the risk assessment, chasing the owner for the documentation that should already exist, and keeping the AI inventory honest. Most weeks are reading and writing, not tooling.
What skills do employers test for?
Reading a model card and spotting what is missing. Writing clearly enough that a lawyer and an engineer both accept it. Knowing the NIST AI RMF and ISO/IEC 42001 well enough to cite the control, not the chapter.
How do people get into this role?
Privacy analysts and internal auditors move in fastest, because the job is evidence and judgment before it is technology. Engineers often find the writing harder than they expect.
What is it often confused with?
Not a data scientist. The analyst asks what the model is for, who it affects and what evidence exists, and does not build it.
Can an AI governance analyst approve or block an AI system?
Not on their own authority. The analyst prepares the assessment and the recommendation. Approval, risk acceptance and any decision to stop a deployment sit with the system owner, a delegated approver or the AI governance committee, as the organization's charter assigns them.