aiqbench analysis
This paper argues that standard technical audits, which rely on aggregate metrics, can miss identity-specific harms to queer communities, such as erasure and misclassification. Its participatory workshop suggests that involving affected people can trace harms to problem definition and data collection, and can expose contract terms that extract value from vulnerable users. The study is qualitative and small (16 participants, four cases), so its value lies in a method for evaluation teams to test rather than in generalizable findings.
Queering the Audits: Community-Based Auditing of AI Harms to Queer Communities
arXiv:2611.00345v1 Announce Type: new Abstract: AI systems embed majority-group defaults into training data, evaluation metrics, and category definitions, producing documented harms for queer communities including erasure, misclassification, and discrimination. Standard technical audits often rely on aggregate measures and cannot detect harms that be come visible only through the lived experience of affected communities. We conducted a participatory auditing workshop at EurIPS 2025 where 16 queer community members audited four case studies using the 4Cs harm taxonomy (Content, Conduct, Contact, Contract) applied across the AI lifecycle. Participants used structured worksheets and plenary synthesis to classify harms and trace them to their origins in the development pipeline. Across all four cases, participants traced harms to problem definition and data collection, and they identified contractual structures that extract value from vulnerable populations while providing minimal recourse. These findings illustrate that community-informed auditing surfaces concrete, identity-specific harms that aggregate evaluation methods risk overlooking.
Abstract:AI systems embed majority-group defaults into training data, evaluation metrics, and category definitions, producing documented harms for queer communities including erasure, misclassification, and discrimination. Standard technical audits often rely on aggregate measures and cannot detect harms that be come visible only through the lived experience of affected communities. We conducted a participatory auditing workshop at EurIPS 2025 where 16 queer community members audited four case studies using the 4Cs harm taxonomy (Content, Conduct, Contact, Contract) applied across the AI lifecycle. Participants used structured worksheets and plenary synthesis to classify harms and trace them to their origins in the development pipeline. Across all four cases, participants traced harms to problem definition and data collection, and they identified contractual structures that extract value from vulnerable populations while providing minimal recourse. These findings illustrate that community-informed auditing surfaces concrete, identity-specific harms that aggregate evaluation methods risk overlooking.
Computation and Language (cs.CL); Computers and Society (cs.CY)
arXiv:2611.00345 [cs.CL]
(or arXiv:2611.00345v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2611.00345 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
In Proceedings of the Second Workshop of Identity Aware AI at 2026, pages 66-76, Palma de Mallorca, Spain. European Language Resources Association
https://doi.org/10.63317/5acrwqmesyw4 Focus to learn more DOI(s) linking to related resources
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 275 words