Yes, AI hiring tools can discriminate against disabled applicants if they screen out qualified people because of a disability, an accommodation request, a medical disclosure, or data that quietly stands in for disability. That risk shows up in resume filters, assessment tests, automated interviews, and other screening steps that look neutral on paper. The issue is getting fresh attention as courts let key claims in the Workday AI hiring litigation move forward, which has pushed algorithmic fairness back into the conversation.
Employers can still face serious liability even when a third party runs their hiring pipeline. Courts have made clear that adopting a vendor's technology does not hand off a company's duty to provide equal employment opportunity. So it pays to understand how these tools actually work, whether you're screening candidates or applying for the job yourself.
Why AI Hiring Tools Can Create Disability Discrimination Risks
AI tools do not need to "know" someone's diagnosis to create discrimination risk. These systems lean on historical training data, rigid productivity scoring, or communication analysis that can quietly penalize qualified candidates. The risk usually traces back to two ideas: intentional discrimination and disparate impact.
Intentional discrimination happens when someone designs or uses a tool specifically to weed out candidates based on disability or accommodation-related information. Disparate impact is different. A neutral-looking system screens out disabled applicants at a much higher rate without sufficient business necessity. Both theories carry weight in compliance reviews.
Courts have allowed significant claims involving AI-driven hiring bias tied to disability, race, age, and sex to proceed in recent litigation. The Workday case matters because it tests whether software vendors and employers can both face scrutiny when automated tools shape outcomes. Workday has denied the allegations, stating its technology focuses on job qualifications rather than protected traits.
The stakes for the job market are hard to overstate. More than 80% of U.S. employers and virtually all Fortune 500 companies reportedly use AI tools in hiring. One recent report put the figure at 73% of employers now using AI in hiring decisions. About half of those employers say their tools automatically reject up to 50% of applications before a human ever looks.
Where Disability Bias Can Show Up in the Hiring Process
Disability discrimination in AI hiring rarely looks like a blunt question about medical history. The bias usually hides inside the data points the software rewards or punishes. HR teams and applicants should keep a close eye on several specific stages.
Automated Resume Filters
Resume parsing software scans documents for keywords, work history, and formatting. Employment gaps can trigger automatic rejection, even when those gaps trace back to medical treatment, recovery, or disability leave. An unusual career path might reflect accommodation needs or an episodic disability, yet the algorithm may score it poorly anyway.
Keyword screening also tends to undervalue nontraditional experience. And applicants who use alternative resume formatting or speech-to-text software may run into parsing accuracy issues. If the AI can't read the text cleanly, it may reject the applicant regardless of what they can actually do.
Pre-Employment Assessments and Timed Tests
Plenty of companies use timed cognitive, personality, or game-based tests to measure aptitude. These often disadvantage people with visual, motor, neurological, psychiatric, or learning disabilities. Picture a candidate with a motor impairment scoring low on a reaction-time game that has nothing to do with the actual job.
If an assessment platform isn't accessible, the problem sits with the tool's design, not the applicant's qualifications. Employers still need a real, reasonable accommodation process for assessments. Companies have to rethink how they screen people so third-party tests don't unlawfully filter out capable workers.
Video Interviews and Communication Analysis
Some automated systems score video interviews on eye contact, tone of voice, facial expressions, pauses, and speech patterns. That scoring can unfairly penalize disabled applicants who communicate differently. A system trained on "typical" facial expressions may wrongly flag an excellent candidate as unenthusiastic or nervous.
This technology is especially risky for applicants with autism, speech differences, facial paralysis, hearing impairments, anxiety, PTSD, or neurological conditions. When a machine grades human communication without context, what does it default to? Too often, a discriminatory pattern.
Background Checks and Medical-Adjacent Data
A standard background check is separate from disability status, but risk creeps in when hiring systems ingest extra data. Algorithms might infer medical history from workers' compensation records, employment gaps, benefit-related data, or health-related disclosures found online. That data can effectively act as a proxy for disability.
Employers should steer clear of collecting or using data that stands in for medical conditions. Unless it's legally justified and directly job-related, feeding it into a predictive hiring model invites significant legal exposure.
Accommodation Requests During Hiring
Candidates often need to request extra time, a screen reader-compatible format, captioning, or interpreter support during hiring. Asking for an alternate interview format or a schedule tweak should never register as a negative signal in an applicant tracking system.
Yet some systems may inadvertently flag candidates who step outside a standard workflow to request an accommodation. Retaliation or exclusion after such a request remains a major risk area that employers need to watch actively.
Comparison Table: Intentional Discrimination vs. Disparate Impact
Understanding how AI tools run afoul of the law means knowing the gap between direct bias and unintended consequences. The table below breaks down the two primary legal theories.
| Issue | Intentional Discrimination | Disparate Impact |
| What it means | The tool or decision-maker treats disability as a reason to reject or downgrade a candidate. | A neutral tool or rule disproportionately excludes disabled applicants. |
| Example | Recruiters treat accommodation requests in the system as a red flag for future performance. | A timed math test automatically screens out applicants who simply need extra time to finish it. |
| Proof focus | Direct evidence, communications, design choices, and policy use. | Statistical patterns, screening outcomes, job-relatedness, and business necessity. |
| HR response | Stop the practice immediately and investigate the people involved. | Audit outcomes, validate job relevance, and add accommodations and human review. |
Both theories matter in compliance reviews and litigation. A company doesn't need to act with malice to break the law; failing to test a vendor's algorithm for disparate impact can create the same legal exposure.
What the EEOC and HR Teams Should Pay Attention To
Employers can still be on the hook when they use third-party AI or screening vendors. The "the vendor built it" defense doesn't fully protect a business. Equal Employment Opportunity Commission (EEOC) guidance stresses that screening criteria must be job-related and consistent with business necessity.
The scale here is significant. In fiscal year 2022, there were 1,338 disability discrimination charges filed in California with the EEOC, making up 30.4% of all discrimination charges in the state. And in 2023, disability discrimination was the most common basis for employment complaints filed with the California Civil Rights Department, accounting for 13,686 complaints.
Practical steps for HR teams:
● Audit automated screening rules before deployment to catch biased proxies early.
● Test whether assessment tools work on common assistive technologies like screen readers.
● Offer a clear accommodation request path before interviews or tests begin.
● Require vendors to explain what data the tool uses and what it explicitly ignores.
● Build in human review for rejections triggered by rigid software thresholds.
● Track complaints and rejection patterns for signs of disparate impact.
● Revalidate selection criteria when job requirements change, so business necessity still holds up.
Human review should always exist for borderline candidates or anyone a machine rejects automatically. Reviewing vendor contracts for testing rights, audit rights, and indemnification language is an essential step for HR leadership.
What Disabled Applicants Should Document If They Suspect AI Bias
Applicants rarely get to see the algorithm itself, which makes personal documentation vital. If you suspect an AI hiring tool screened you out unfairly, preserve facts that show timing, disclosure, and inconsistency. Start by saving the original job posting, the exact date you applied, and the specific version of the resume you submitted.
Save the test instructions, the platform name, and screenshots of any accessibility barriers or error messages too. If you requested an accommodation, keep copies of those emails or web forms. Watch the timing of your rejection closely, especially if it lands right after a medical disclosure or an accommodation request.
For California readers, this rundown of the 7 signs an employer is discriminating based on disability in California can help turn a vague hunch into specific warning signs worth documenting.
Where you can, keep records of the qualifications that appear to match the role. Repeated rejections through the same automated system for jobs you're clearly qualified for can point to a systemic barrier. And keep this documentation outside employer-controlled systems, so you don't lose access to it.
What This Means Going Forward
AI doesn't remove discrimination risk; it can scale it fast. Employers need to treat AI screening as a critical piece of hiring, not an unquestionable black box. Disability bias tends to surface through design choices, inaccessible assessments, or proxy signals rather than explicit labels.
AI can speed up hiring, but it doesn't cancel disability rights. Employers should audit their tools proactively, and applicants who spot a troubling pattern should document it early. Good HR practice ultimately comes down to accessibility, auditability, and meaningful human review.
