{"paper":{"title":"Position: Age Estimation Models Do Not Process Biometric Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Age estimation models do not process biometric data because they cannot identify individuals.","cross_cats":["cs.CV","cs.LG"],"primary_cat":"cs.CY","authors_text":"Nikita Marshalkin","submitted_at":"2026-05-17T09:37:28Z","abstract_excerpt":"When a neural network estimates someone's age from a photograph, does it process biometric data? The answer depends on whether identity-discriminative representations arise within the network during inference, a question that may seem trivial to ML researchers but triggers consent requirements under GDPR, statutory damages under BIPA, or high-risk AI classification under the EU AI Act. Yet no regulatory guidance addresses it. This position paper provides empirical evidence: 14 models evaluated across 3 face verification benchmarks show age estimators fall orders of magnitude short of identific"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"14 models evaluated across 3 face verification benchmarks show age estimators fall orders of magnitude short of identification thresholds. Age estimation models cannot identify individuals.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the chosen face verification benchmarks are sufficient to detect whether any identity-discriminative representations arise inside the age estimation network during inference, and that falling short of identification thresholds equates to not processing biometric data under GDPR/BIPA/EU AI Act definitions.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Empirical evaluation shows age estimation models perform orders of magnitude below identification thresholds on face verification benchmarks, indicating they do not extract identity-discriminative representations.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Age estimation models do not process biometric data because they cannot identify individuals.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"8e5886b19b25f2afe3360036ba51c37cfa874eb4c0cf1fee65c14f932b0d7512"},"source":{"id":"2605.17347","kind":"arxiv","version":1},"verdict":{"id":"272f576a-01af-4690-9b90-2e34df2a5908","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T23:19:31.485509Z","strongest_claim":"14 models evaluated across 3 face verification benchmarks show age estimators fall orders of magnitude short of identification thresholds. Age estimation models cannot identify individuals.","one_line_summary":"Empirical evaluation shows age estimation models perform orders of magnitude below identification thresholds on face verification benchmarks, indicating they do not extract identity-discriminative representations.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the chosen face verification benchmarks are sufficient to detect whether any identity-discriminative representations arise inside the age estimation network during inference, and that falling short of identification thresholds equates to not processing biometric data under GDPR/BIPA/EU AI Act definitions.","pith_extraction_headline":"Age estimation models do not process biometric data because they cannot identify individuals."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.17347/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T23:31:20.101265Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T23:31:00.586635Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T21:41:57.796996Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T21:33:23.733558Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"746ee6a77d03cabf1d03c9131e11d987f2a74dd7d03b0e82822fb2c0795a823d"},"references":{"count":40,"sample":[{"doi":"10.1109/iccvw54120.2021.00166","year":2021,"title":"Llvip: A visible-infrared paired dataset for low-light vision","work_id":"d4e2d955-81eb-4270-a9ba-e18548695985","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2012,"title":"Opinion 3/2012 on developments in biometric technologies","work_id":"9edb3fc2-3ae1-4eb5-8c66-ab31616a74fd","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"FaceLib : Detection, age gender estimation & recognition","work_id":"b75d489b-0c29-4645-8c35-a5ae4547b198","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2026,"title":"H., Madotto, A., Wei, C., Ma, T., Zhi, J., Rajasegaran, J., Rasheed, H","work_id":"aab442d9-7b48-491c-bf07-4610cd052139","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Cam\\'eras augment\\'ees pour estimer l'\\^age dans les bureaux de tabac : la CNIL pr\\'ecise sa position","work_id":"c3df3fd7-40f4-4d7b-a9e1-fe5339d12ca8","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":40,"snapshot_sha256":"3a3da7924e38511ddfc5a661c1c7e3d5bb0a3d35a6876adda0eedc8d4d190642","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}