9. AI in People Work

Governance gates for AI touching employees

The pre-pilot checklist

Use when

Before any pilot that processes employee data or affects work

Governance gates for AI touching employees, shown as a sequence of 5 parts: Privacy assessment, Bias test, Consultation check, Named human, Audit trail.
  1. 1

    Privacy assessment

    PIA against applicable law (PIPA/PIPEDA); data minimized

  2. 2

    Bias test

    Documented, repeated after model changes

  3. 3

    Consultation check

    Union/works-council duties (e.g. BC LRC s.54) and CA clauses

  4. 4

    Named human

    Accountable owner per workflow; outputs are inputs

  5. 5

    Audit trail

    Logged decisions; grievable exactly as before

How to use it

Run the five gates as a standard checklist per use case; partner with privacy, legal, and the enterprise AI team rather than building shadow governance. The statutory consultation gate varies by jurisdiction; in BC, s.54 requires 60 days notice plus adjustment-plan effort for tech change affecting a significant number of unionized employees. Verify current law each time; this is a checklist, not legal advice.

Card details

Model type
Sequence
Origin status
Synthesis
Content last reviewed
28 July 2026

Origin

Synthesis

Standard privacy and AI governance practice, assembled here for people processes. The specific additions are the statutory consultation gate, including BC Labour Relations Code s.54, and the requirement of a named accountable human per workflow. Assembled by Harsimran Kaur Kapoor, 2026. Statutory requirements vary by jurisdiction and must be verified.

Neighbouring models

NextNow, next, later

Was this useful?