Automated employment decisions, audit-ready.
NYC Local Law 144 requires bias audits of automated employment decision tools. The EU AI Act classifies hiring AI as high-risk. Both regulations demand the same thing: a per-decision record an auditor can read.
Page in development
Concrete export samples, auditor-access flow walk-through, and LL 144 pack format detail will land on this page as we ship.
The shape of the problem.
AI used to screen résumés, rank candidates, or otherwise narrow the hiring funnel. The buyer is the head of people-ops or the GC. The jobs are (a) commission an LL 144 bias audit annually, and (b) maintain EU AI Act Annex IV technical documentation if the system is deployed in the EU.
What Trustvice produces.
- Per-decision record — input class, output, model version, timestamp, and the rule-set evaluated.
- Dissent log — every case where the agent's output disagreed with one of your stated equal-treatment rules, even if it proceeded.
- LL 144 pack export — the data your chosen audit firm needs, in the format they ask for.
- Annex IV companion document — populated from the operational record (deployment context, monitoring procedure, training inputs).
- Auditor read-only scope — issue a workspace-scoped, read-only token to your audit firm, with full audit-log visibility on what they viewed.
What lives elsewhere.
The bias-audit calculation itself. The hiring criteria. The remediation decision when an audit reveals a problem. Those are the deployer's responsibility — Trustvice produces inputs to that work, not outputs.
Trustvice 360