Public AI Use Policy

Plumb uses AI for drafting and summarising: job descriptions, application summaries, interview question suggestions, rejection feedback drafts, translation, and process summaries. Every AI output is reviewable and editable by a human.

Regardless of which features an organisation switches on, Plumb does not use AI to:

• make or finalise a hiring decision
• reject, advance, or invite a candidate
• filter, screen out, or auto-shortlist anyone
• apply a cut-off, pass mark, or threshold
• see or use demographic data for any purpose

Every candidate stays visible to the people running the hire, and a person reviews and decides each one. Hiring decisions stay with people.

An organisation can choose to switch on advisory AI ranking. When it is on, AI suggests an order for the review queue based on the role's own scoring rubric, alongside a confidence level and a trace of the evidence it used. It is a starting order, not a verdict.

Advisory ranking is bound by clear limits:

• It only orders the queue — it never rejects, advances, filters, or shortlists anyone, and never applies a cut-off.
• Every candidate remains visible, and a person reviews and decides each one.
• Reviewers can re-sort to a neutral, non-AI order at any time.
• It runs only on anonymised application evidence and the role rubric — never on names, contact details, or demographic data.
• Every ranking run is logged and auditable.

Ranking is off unless an organisation deliberately turns it on. Where it is used, the organisation discloses that, including to candidates where the law requires notice.

Candidates are not penalised for using AI to prepare, understand a question, or improve clarity.

Where an organisation has switched on advisory AI ranking for a role, candidates are told that AI is used to help order applications for human review, and that a person decides every outcome.

EU AI Act: recruitment and candidate-evaluation AI can fall within high-risk employment uses. Plumb is designed around human oversight, transparency, audit logs, and no automated selection — advisory ranking orders a queue for people to review and never decides.

NYC Local Law 144: where an organisation enables advisory ranking, it remains responsible for the applicable bias-audit and candidate-notice obligations. Plumb keeps ranking off by default, provides the audit trail and disclosure to support those obligations, and keeps human discretion central — AI does not substantially replace hiring judgement.

Colorado AI Act: Plumb keeps candidate-affecting decisions with the employer and keeps records that support notices, reviews, and impact-assessment work.

Illinois: Plumb does not allow AI to make discriminatory employment decisions, and AI use is visible rather than hidden.

AI use in Plumb is visible rather than hidden, and AI actions are logged so they can be reviewed. Where advisory ranking is used, its use is recorded and surfaced — including, where a public Hiring Receipt is published, a note that AI ranking assisted the review.