AI Bias & Fairness

This page provides transparency about the fairness testing of Wigiwork's automated matching, ranking, and shortlisting systems. It is intended to support our obligations under the EU AI Act (Regulation (EU) 2024/1689), NYC Local Law 144 where applicable, and comparable equality and anti-discrimination laws. It is a summary for users and is not a substitute for the full technical documentation maintained internally.

This audit summary covers the automated systems that influence which anonymous candidates are surfaced to employers, including:

• Candidate-job matching and relevance scoring
• Discovery Search ordering
• Automated shortlisting and inclusion
• Profile Strength as it affects ordering or visibility

It does not cover the Wigi AI Assistant's conversational outputs, which do not rank or select candidates.

Our testing examines whether automated outcomes differ unjustifiably across protected and sensitive characteristics, including:

• Sex and gender
• Race and ethnicity
• Age (beyond lawful eligibility checks)
• Disability
• Nationality (beyond lawful compliance filters)

We assess outcomes such as shortlist inclusion rates and ranking position to identify disparities that are not explained by legitimate, role-related professional factors.

Wigiwork applies a combination of methods designed to detect and reduce bias:

• Statistical analysis of selection and ranking outcomes across groups, including disparate-impact ratios
• Review of input features to confirm identity and protected attributes are excluded from ranking signals
• Testing on representative samples rather than identifiable individuals
• Human review of flagged disparities by trained personnel
• Documentation of findings, mitigations, and re-testing

By design, Wigiwork's systems operate only on professional and platform-related information and do not use identity or protected demographic attributes as ranking factors.

Wigiwork reviews these systems for bias and fairness:

• Before a material new model or ranking change is deployed
• At least annually thereafter
• On an ad-hoc basis where monitoring or a complaint indicates a potential issue

The most recent review date is shown in the "Last updated" field above.

Wigiwork conducts this testing proactively, ahead of the 2 December 2027 applicable date for standalone high-risk obligations under Annex III of the EU AI Act, as part of our ongoing alignment program described in our AI Act Compliance statement.

Where required by applicable law, bias audits are conducted or validated by a suitably qualified independent party. Internal monitoring is performed by personnel who are not solely responsible for the design of the systems being tested, to support objectivity.

Where Wigiwork's tools are used as an "automated employment decision tool" for positions located in New York City, Wigiwork supports the additional requirements of NYC Local Law 144, including:

• An independent bias audit within the preceding year
• Publication of a summary of audit results, including impact ratios by sex, race/ethnicity, and intersectional categories
• Candidate notice of the use of automated tools

Where these obligations apply, the corresponding published audit results will be made available on or linked from this page.

Wigiwork's automated systems support discovery and shortlisting but do not make hiring decisions, which rest solely with employers. Employers remain responsible for ensuring their use of recommendations, shortlists, and rankings complies with applicable anti-discrimination and employment laws. Wigiwork is not an employer or recruitment agency and does not make employment decisions.

No fairness testing method is perfect. Audits reduce, but cannot fully eliminate, the risk of bias, and results reflect the data and methods available at the time of testing.
Outcomes may change as models, data, and job criteria evolve, and this summary will be updated to reflect material changes.

Where applicable law provides rights related to automated processing, you may request information about how automated tools were used, ask for human review, object to automated processing, or request correction of inaccurate AI-derived data. To exercise these rights, contact privacy@wigiwork.com as described in our AI & Automated Decision Notice and Data Protection Notice.

For questions about this summary or our fairness testing:

Wigiwork Inc. PBC
Postal: 2261 Market St, San Francisco, CA 94114-1612, USA, Attn: Legal
Email: legal@wigiwork.com (general) · privacy@wigiwork.com (data rights)