AI governance: AI Settings per area, AI Decision Reviews, the AI-use report and the AI Disclosure Register
Set, area by area, whether AI and automated decision tools may act — off, suggest-only or allowed — enforced where each tool runs: ASK, agents, resume scoring and AI summaries. Under suggest-only every AI output about a person opens a review a person decides. Download the AI-use report and the disclosure register for an audit or an RFP.
どこで見つけられるか
- Platform › Access & Security › AI Governance › AI Settings, AI Decision Reviews, AI Use and AI Disclosure Register.
- AI Decision Reviews also sit under HRMS › Compliance › Employee Relations & Records (HR areas and recruiting) and Talent & Contingent Workforce › Talent & Pipeline › Activity (recruiting).
- Roles: Security & AI Governance Admin sets AI Settings and decides reviews; Governance Auditor (read only) sees all of it; Human Resources, Recruitment Coordinator and Recruiter decide the reviews of their area.
AI Settings
Since 22 October a new company starts with every area allowed — AI Settings says “AI features are allowed by default for this company; an administrator can turn any area off.” Turn an area off, or to suggest, below. Companies that existed before keep their settings.
- Pick the area’s row and choose Set.
- Under AI and automated decision tools choose off, suggest or allowed, and add a Note — the policy or board decision behind it, for example <HR policy 4.2, 12 October 2026>.
- The setting applies at once, everywhere the tool runs, not just on a screen.
| 設定 | その機能 |
|---|---|
| オフ | The tools do not run, and ASK and agents cannot read or change the area’s records: “AI is switched off for <Performance> in this company (Platform → AI Governance), so this cannot be read or changed through AI. A person does it on the screen.” |
| suggest | They may propose and a person decides: every AI output about a person opens a review, ASK proposes but does not change the area, and an agent’s change waits for approval. |
| allowed | As the product behaved before. Every area starts as allowed. |
The areas are fixed: Recruiting: screening and ranking, Onboarding, Performance, Engagement, Workforce analytics, Compensation, Promotion, Discipline, Termination, Training recommendations, Succession, ASK assistant (general) and Agentic AI agents (general). The Employment decision column marks those that decide about a person. An area cannot be removed; set it off instead.
AI Decision Reviews
- A review (AIR000001 and up) opens by itself when resume scoring or a candidate summary is made while its area is suggest-only.
- Choose Decide on a pending review. Decision: accepted (the output stands), overridden (say What you decided instead) or rejected (the output is not used).
- Give Why — required unless you accept. The decision is recorded with your name and the reason.
The AI-use report and the disclosure register
- AI Use lists every use of AI in the last 90 days, and every refusal: ASK questions, agent runs, resume scoring, AI summaries and support auto-answers, with the Outcome ran, suggested or refused. Download the AI-use report asks for Days back (default <365>) and gives a CSV of uses and a summary.
- AI Disclosure Register describes each tool — resume screening, the AI candidate summary, the AI job order summary, ASK, agents, the support auto-answer and interview panel scoring — with how it is set here, the data it uses and how a person reviews it. Download the register gives a CSV and a JSON.
- Company data is not used to train AI models. Both downloads are audited.
How to reuse: the area, note and number of days in angle brackets are examples; choose your own.