domain
Employment Law
High-risk AI territory under the EU AI Act, handled as engineering
Employment is a named high-risk category under the EU AI Act, which makes it one of the clearest places to see what CueCrux is for. If agents assist with screening, hiring analytics, or workforce decisions, the deployer needs demonstrable human oversight, not a paragraph in a policy asserting it exists.
In CueCrux, plans and work items carry risk classes, and high-risk work routes through human gates that cannot auto-approve past a timeout. Each approval is recorded as a signed fact; each policy refusal fails closed with a signed RefusalReceipt carrying a reason code. When an agent declines to act on a candidate record, that refusal is itself evidence of the control operating.
Hash-chained logging covers the record-keeping and traceability expectations of the Act (Articles 9, 10, 12, 13, 14, 15, and 50 are the relevant machinery mappings), and GDPR Article 17 erasure is handled by scoped forget with a dry-run preview. These are engineering controls, not a legal opinion, and that is the point: when asked to demonstrate oversight, you replay it.
For the underlying legal research, cross-jurisdiction employment questions resolve against cited sources, so the answers your team acts on open their evidence.