Govern · agent risk record · Human Resources

Screening Agent

depth-agent

All agents

Risk class

High risk

EU AI Act Annex III(4)(a)

Status

Operating with a gap

Oversight

Recruiting pod lead, per requisition

Recruiter

Explanation

Published

Candidate notice, published in the application flow in five languages

Appeal path

Named human

120h service level

Conformity

Self-assessed

assessed 41d ago

Declared gap

Self-assessed. No notified body involvement, and the NYC Local Law 144 independent bias audit has not been commissioned even though the agent screens candidates for New York roles.

Classification

Why this agent sits in this class, under which regime

Regime

EU AI Act Annex III(4)(a)

Class

High risk — Affects access to employment, credit or a comparable material interest. Requires human oversight, logging and contestability.

Basis

Employment: a system used to filter applications and evaluate candidates in the course of recruitment. High risk on the face of Annex III.

People affecting

Yes — the output changes an outcome for a named person

Scope

human-resources · human-resources/talent-acquisition

Registered

268d ago

Log retention

84 months of decision logs retained

Conformity

Self-assessed — Assessed by the operator. Not independent.

Human oversight

The measure, and the person it is attached to

Measure

Recruiter review of every shortlist, with the ranking rationale and the field list visible, and the ability to reinstate any applicant.

Named human

Recruiting pod lead, per requisition · Recruiter

Autonomy

Transparency notice

Served — the affected person is told an automated system is involved

Explanation surface

Published — A standing explanation is available to the affected person without asking.

Explanation reference

Candidate notice, published in the application flow in five languages

How an affected person contests this

The route, the named human at the end of it, and the service level

5 appeals on record

Route

Talent Acquisition Lead for the region

Path type

Named human

The appeal lands with a person, not a queue.

Service level

120 hours

1 appeal past that service level right now.

Reference	Subject	Grounds	Routed to	Age	Review	Outcome	Evidence
APL-0011 Past service level	CAND-49512 candidate · US · Candidate portal	Applicant requests human review of an automated ranking under state law.	Ben Ortiz Talent Acquisition Lead, Americas	168h of 120h	— human time	Pending	HR2A-1021 #3
APL-0009 Closed	CAND-49330 candidate · AU · Candidate portal	Applicant states an accessibility adjustment was not accounted for in the assessment stage.	Lin Wei Talent Acquisition Lead, APAC	62h of 120h	44m human time	Overturned The adjustment request had not reached the assessment stage. Reassessed with the adjustment in place and shortlisted.	HR2A-1008 #2
APL-0006 Closed	CAND-49014 candidate · SG · Candidate portal	Applicant requests the reason for rejection and the data used.	Lin Wei Talent Acquisition Lead, APAC	43h of 120h	19m human time	Original upheld Rationale and field list supplied. No change to the outcome.	HR2A-1002 #2
APL-0002 Closed	CAND-48377 candidate · US · Candidate portal	Applicant asks on what basis the application was ranked and requests human review.	Ben Ortiz Talent Acquisition Lead, Americas	96h of 120h	22m human time	Original upheld Ranking rationale supplied in full. Human reviewer reached the same conclusion on the stated requirements.	HR2A-1021 #2
APL-0001 Closed	CAND-48120 candidate · UK · Candidate portal	Applicant states that relevant experience held under a former job title was not recognized by the ranking.	Aisha Bello Talent Acquisition Lead, EMEA	71h of 120h	38m human time	Overturned The former job title was not in the skills taxonomy. The applicant was added to the shortlist and the taxonomy was extended.	HR3B-1010 #1

Adverse-impact readings behind this agent

Candidate screening and shortlist · Aug 2026

Full monitoring record

Monitoring status

Monitored

Population a month

8,420

Review cadence

Monthly cohort review, quarterly threshold review

Nadia Kovac, Data Protection Officer, with Priya Raman

Last reviewed

19d ago

Cohort	Compared with	n	Selection rate	Reference rate	Impact ratio	Reading
40 to 49 age-band	30 to 39	1,937	15.5%	18.2%	0.85	Within threshold 40 to 49 is selected at 85 percent of the rate for 30 to 39.
Under 30 age-band	30 to 39	2,273	20.1%	18.4%	1.09	Within threshold Under 30 is selected at 109 percent of the rate for 30 to 39.
50 and over age-band	30 to 39	1,179	14.0%	18.4%	0.76	Outside threshold 50 and over is selected at 76 percent of the rate for 30 to 39.
Minority ethnic groups, aggregated Ethnicity	Majority group (per local reporting standard)	2,610	15.9%	17.9%	0.89	Within threshold Minority ethnic groups, aggregated is selected at 89 percent of the rate for Majority group (per local reporting standard).
Women Gender	Men	3,705	16.0%	18.4%	0.87	Within threshold Women is selected at 87 percent of the rate for Men.

Open alarms on this decision point

Derived from the readings, never authored by hand

FAIR-0001Sustained driftMedium severityAcknowledged

The ratio for Women has moved from 0.93 to 0.87 across 6 periods without ever breaching. This is inside tolerance and heading out of it, which is the reading a threshold alone would miss.

Owner Nadia Kovac · opened 170d ago

FAIR-0002Never reportableLow severityMonitoring

No reading has ever been produced for Non-binary and self-described on this decision. The cohort is declared in scope and the population exists, but it has never once been large enough to report against Men at the confidence this program requires.

Owner Nadia Kovac · opened 170d ago

FAIR-0003Sustained driftMedium severityAcknowledged

The ratio for 40 to 49 has moved from 0.91 to 0.85 across 6 periods without ever breaching. This is inside tolerance and heading out of it, which is the reading a threshold alone would miss.

Owner Nadia Kovac · opened 170d ago

FAIR-0004Threshold breachHigh severityOpen

50 and over is being selected at 76 percent of the rate for 30 to 39, below the four-fifths threshold. The breach has held for 3 of the six periods.

Owner Priya Raman · opened 78d ago

What this record is, and what it is not

Read this before quoting anything above

Fairness readings computed on modeled data demonstrate the instrument, not a real audit result. The classification, the oversight measure, the explanation surface and the appeal route are design decisions recorded here so they can be argued with; the population behind the readings is modeled.

No agent on this register carries an independent third-party conformity assessment. Every assessment on file was made by the operator.

Appeal outcomes are recorded against the evidence spine where a unit reference exists. Where the evidence column reads “not recorded”, the appeal was resolved outside the spine and that is a gap, not a rounding.