Human accountability
Material candidate/employment decisions remain attributable to an accountable human rather than hidden behind model output.
Recruitment technology governance
Responsible Recruitment AI Standard
A practical governance standard for recruitment agencies and employers using AI in sourcing, screening, communications, scheduling, CRM/ATS workflows and decision support. It is not a legal certification.
Purpose limitation
AI use has a defined recruitment purpose and is not expanded into unrelated profiling merely because data is available.
Data minimisation
Candidate and employee data is limited to what is necessary, authorised and appropriate for the task.
Bias & performance
High-impact models/workflows are evaluated for task performance and potentially harmful differential effects using suitable evidence.
Explainability & challenge
Where consequential automation affects people, the organisation should be able to explain the process sufficiently for governance and appropriate challenge.
Vendor diligence
Recruitment AI suppliers are assessed for data use, retention, security, model changes, subprocessors and contractual controls.
Automation boundaries
Outbound messages, scheduling, ranking and workflow actions use approval/escalation rules proportionate to consequence.
Auditability
Material inputs, model/workflow versions, decisions and human overrides should be traceable where appropriate.
No fake human
AI-assisted candidate communication should not be designed to deceptively impersonate a real named recruiter.
Review & retirement
AI workflows are reviewed for continued value, accuracy, fairness, complaints and changed legal/operating context.