Recruitment AI can affect access to employment before a human decision-maker ever meets the candidate. Under Annex III, systems used to target job ads, filter applications or evaluate candidates can be high-risk. Interview tools add another boundary: if they infer emotions in the workplace, Article 5 can prohibit the use outright.
The Regulation lists specified recruitment and candidate-evaluation uses as high-risk and prohibits workplace emotion inference except for medical or safety reasons. Current Commission guidance states that Annex III employment high-risk rules apply from 2 December 2027.
Start with the actual system, role, use case and evidence boundary. The same regulation can produce different obligations for different actors and systems.
Annex III treats specified employment and recruitment systems as high-risk, including systems used to place targeted job advertisements, analyse or filter job applications, and evaluate candidates.
Article 5 prohibits AI systems used to infer emotions of natural persons in the workplace, except for medical or safety reasons. A hiring or interview context should therefore be screened for prohibited emotion-recognition features before high-risk controls are considered.
High-risk recruitment AI requires effective human oversight. Preserve who reviews recommendations, what information they receive, when they can disregard or reverse outputs, and how the final employment decision remains attributable.
Preserve system and model version, intended purpose, application inputs, scoring or ranking outputs, interview data, human review, overrides, decision criteria, notices, complaints or challenges, and the evidence supporting the final decision.
The goal is not merely to reach an answer. It is to preserve what facts, evidence, scope and limitations supported that answer at that time.
Separate sourcing, targeted advertising, CV screening, ranking, interview transcription, interview scoring, skills assessment, background analysis, scheduling and final decision support. Not every recruiting tool performs the same regulated function.
If the system infers emotions from biometric data in the workplace or recruitment process, test the Article 5 prohibition first. A prohibited use cannot be made permissible merely by adding high-risk controls.
Connect input data, scoring logic, system output, human reviewer, decision authority, override, notices, challenge pathways, logs, bias or performance evidence and limitations to the consequential hiring outcome.
A new scoring model, interview-analysis feature, changed job criteria, new data source, broader automation, removed human checkpoint or different employment use can make the previous classification and evidence stale.
Choose the smallest operating tier that fits the portfolio today. Upgrade when system count, team size or governance scope actually requires it.
Keep a living system-level evidence record with obligations, gaps, versions and revalidation state.
START EVIDENCE PASSPORT →Coordinate evidence, owners, documentation, incidents and team compliance work in one governed workspace.
START COMPLIANCE WORKSPACE →Operate broader high-risk, GPAI, FRIA, post-market and material-change governance across a growing portfolio.
START GOVERNANCE PRO →Run institutional governance with expanded users, authority workflows, examiner rooms and portfolio reporting.
START INSTITUTION →Use the free classifier to establish the system, intended purpose, possible actor role, EU exposure and unresolved facts. When continuing evidence infrastructure is needed, paid access begins at $19 per month. Independent human readiness review remains a separate service.
Certain recruitment systems are. Annex III identifies AI used for recruitment or selection, targeted job advertising, analysing and filtering job applications, and evaluating candidates as high-risk use cases.
No. The legal route depends on what the system actually does. A video interview tool that records or transcribes may differ from a tool that scores candidates. If it infers emotions from biometric data in the workplace, Article 5 prohibits that use except for medical or safety reasons.
High-risk AI must be designed and used with effective human oversight. The responsible human should be able to understand relevant limitations, monitor the system, avoid automation bias, and disregard, override or reverse outputs where appropriate.
Current European Commission guidance states that the high-risk rules for Annex III areas including employment apply from 2 December 2027.
Preserve system identity and version, intended purpose, provider and deployer roles, job and candidate scope, inputs, scoring or ranking outputs, human review, overrides, notices, logs, performance and bias evidence, complaints, changes and revalidation history.
No. TA-14 can preserve the classification basis, prohibited-practice analysis, evidence, human-oversight state, gaps, decision history and revalidation record. It does not provide legal advice, certification, conformity assessment or regulatory approval.
TA-14 Exchange Activity
Live cumulative activity recorded across the public Exchange surface.
···
Visitors
Recorded public visitors
···
Page Views
Recorded Exchange views