AI & HRMay 19, 2026 11 min read
AI interviews are biased in ways human interviewers never were. Here's the 2026 audit data.
Vendors sold AI screening as the cure for human bias. The 2026 audits show it introduced four new biases humans don't share — and amplified two existing ones.
The pitch was beautiful: remove the human, remove the bias. Three years and a wave of NYC Local Law 144 audits later, the picture is less flattering. AI screening tools don't eliminate bias — they redistribute it, and in some cases manufacture entirely new forms of it that no human interviewer would ever exhibit.
11%
lower pass rate for candidates with non-native accents in voice-based AI interviews
NYC LL144 audit aggregate, 2026
9%
lower scores for candidates over 50, controlling for content of answers
EEOC AI Fairness study, 2025
23%
of AI screeners penalized candidates for 'low video engagement' — disproportionately neurodivergent applicants
Open MIC, 2025
3.1×
more variance in scoring the same candidate on two consecutive AI interviews vs. human panels
MIT CSAIL replication, 2026
The four new biases AI introduced
- Accent bias at the acoustic feature layer — humans adjust, models penalize.
- Background visual bias — candidates in poorly-lit home environments score lower on 'professionalism' regardless of answer content.
- Engagement bias against neurodivergent candidates whose eye contact and facial affect don't match neurotypical baselines the model was trained on.
- Run-to-run instability — the same candidate, same answers, scored 0.4 standard deviations apart on two consecutive runs in 38% of cases.
The promise (2022–2024)
- 'Bias-free, structured, repeatable interviewing.'
- 'Eliminates the halo effect.'
- 'Audited annually for fairness.'
- 'Demographic-blind.'
The 2026 audit reality
- Bias measurable across accent, age, neurotype, and lighting.
- Halo effect replaced by 'video aesthetic' effect.
- Audits often run by the vendor on a synthetic dataset, not real production data.
- Models inferred protected characteristics from voice and video with 70%+ accuracy.
What to do before your next renewal
- Demand the actual 4/5ths rule disparate impact data from the last 12 months of production use — not the vendor's synthetic audit.
- Run a same-candidate stability test: have 20 employees take the assessment twice. If scores move more than 10%, the instrument isn't measuring what you think.
- Always pair AI screening with a structured human interview before any rejection decision. Never let the model auto-reject.
- Disclose to candidates that AI is in the loop, what it scores, and how to request a human review. (Required in NYC, IL, CO and the EU in 2026.)
Written by
Pawan Joshi
HR & Operations leader scaling global remote teams across Nepal, the Philippines, Australia, and the US. Tech-leaning writing lives on Medium.