Recruiting sits at the top of the list of HR processes professionals are augmenting with AI, according to Society for Human Resource Management (SHRM) survey respondents. AI touches nearly every stage of the hiring process, from writing job postings to ranking resumes and scoring video interviews.
And while some expect machines to eliminate the bias that has been prevalent in hiring for years, the reality is AI can learn to make the same choices, just on a faster and larger scale. Understanding how AI bias in hiring happens and what the law requires is a core part of implementing AI responsibly.
What Is AI Bias in Hiring?
AI bias in hiring occurs when an algorithm systematically favors or disadvantages candidates based on characteristics like race, gender, age, or ability. Human bias, while unfortunate, affects just one decision at a time. An algorithm’s bias applies the same skewed logic to every resume it touches, meaning a small statistical bias in the underlying model can turn into a large, measurable pattern of exclusion across thousands of applicants.
How Are Employers Using AI in the Hiring Process?
SHRM’s 2026 State of AI in HR shows 39% of organizations have implemented AI somewhere in their HR function, with another 7% planning to do so this year. Recruiting accounts for 27% of that adoption.
AI has moved well past resume keyword matching. From generative AI like chatbots to data analysis and predictive tools, AI supports the hiring process in the following ways:
- Drafting and optimizing job postings
- Screening and ranking applicants
- Scheduling interviews
- Analyzing recorded video responses for tone, word choice, and other signals
- Responding to early candidate questions
- Flagging which applicants are most likely to accept an offer or stay in the role
Why Does AI Bias in Hiring Happen?
Bias enters a hiring algorithm through the data the model learns from, the way people respond to the model’s output, and the indirect signals the model relies on when it can’t see protected characteristics directly.
Biased Training Data
Most AI hiring tools learn from historical hiring data, including past resumes, interview outcomes, and performance ratings. If that history reflects decades of decisions that favored certain candidates, the algorithm will reproduce those patterns at scale.
The Illusion of Objectivity
Because AI output is framed as data-driven, it tends to get less scrutiny than a manager’s gut call, even when it’s built on the same flawed foundation. That perceived objectivity is part of what makes algorithmic bias dangerous.
Proxy Discrimination
AI tools rarely have direct access to protected characteristics like race, gender, or disability status, but they don’t need them to discriminate. Employment gaps, the schools listed on a resume, zip codes, and even affinity group memberships can all correlate closely enough with protected characteristics that a model treats them as reliable stand-ins. The result is discrimination by proxy.
Who Is Most Affected by AI Hiring Bias?
Research across multiple studies points to a consistent set of groups facing disproportionate impact from AI hiring tools. These groups include:
- Women
- People of color
- Older workers
- People with disabilities
- Candidates from non-traditional educational backgrounds
The mechanisms vary. For example, resume screeners trained on historically male-dominated hiring data have downgraded resumes that included the word “women’s,” as in “women’s chess club captain.” Tools that weight continuous work history can penalize candidates with employment gaps tied to caregiving or disability. Models trained on data from a narrow set of feeder schools can systematically undervalue candidates who took a different path into the workforce. In each case, the tool isn’t targeting a protected group directly. It’s optimizing for patterns that happen to track one.
Recent research adds hard numbers to that pattern. A Stanford Graduate School of Business study generated more than 34,500 resumes using a large language model and found it consistently portrayed women as younger and less experienced than men with identical qualifications. When the same model was asked to evaluate resumes, it rated older men most favorably, even when the underlying information matched women’s applications exactly.
That bias doesn’t stay contained to the algorithm. A University of Washington study found that human reviewers don’t just tolerate biased AI recommendations, they adopt them. Without AI screening, the data showed participants’ choices showed little bias. However, when AI surfaced severe biased recommendations, participants followed those recommendations about 90% of the time.
The risk can also spread across an entire industry. A Stanford Digital Economy Lab study of pymetrics-style hiring algorithms found that when many employers rely on the same underlying AI platform, they can end up making similar hiring decisions across companies.
Is AI Bias in Hiring Illegal?
AI laws continue to evolve, but here’s what employers should know.
Existing Federal Protections
Title VII of the Civil Rights Act, the Americans with Disabilities Act, and the Age Discrimination in Employment Act all apply to hiring decisions regardless of whether a human or an algorithm made the call. If an AI tool produces a discriminatory outcome, using a third-party vendor’s software isn’t a defense.
Landmark Cases Employers Should Know
Regulators and courts treat AI-driven discrimination the same way they’d treat a human recruiter who screened out applicants on the same basis.
In the court case Mobley v. Workday, plaintiffs allege that Workday’s AI-powered screening tools discriminate based on race, age, and disability by relying on algorithmic proxies like employment gaps and graduation dates. In July 2024, a federal district court in the Northern District of California ruled that Workday could be sued as an ‘agent’ of the employers using its platform, opening the door for AI vendors to face direct liability under Title VII, the ADEA, and the ADA.
The EEOC has also acted directly against an employer. In its first AI-related hiring settlement, the agency resolved claims against iTutorGroup after its application software was programmed to automatically reject female applicants aged 55 or older and male applicants aged 60 or older. The company paid $365,000 and agreed to years of EEOC monitoring.
The Gap in Federal Legislation
In 2025, the EEOC removed its AI and algorithmic fairness guidance from its website. In 2026, the agency rescinded its FY2024-2028 Strategic Enforcement Plan, which had specifically named AI and algorithmic hiring tools as an enforcement priority, and replaced it with a new National Enforcement Plan for FY2025-2029 that doesn’t mention technology-related discrimination.
None of that changes the underlying law. Title VII and the EEOC’s Uniform Guidelines on Employee Selection Procedures (29 CFR Part 1607) still apply to AI-assisted hiring. What’s missing is federal-level clarity on how AI-specific issues, like proxy discrimination or vendor liability, get evaluated in practice. Until Congress or a future rulemaking addresses it, employers are left interpreting decades-old statutes against tools that didn’t exist when those statutes were written.
State and Local Regulations
With federal guidance in flux, states and cities are filling the gap on their own terms. At the state level, Illinois bars AI-driven employment discrimination, while Colorado and California‘s AI rules are phasing in through 2027. New York City’s Local Law 144 requires bias audits for automated hiring and promotion tools.
This patchwork of laws creates a compliance burden for multi-state companies.
How Employers Can Reduce AI Bias in Hiring
Follow these steps to reduce AI bias in hiring:
Audit AI Tools
Before adding an AI hiring tool to your recruiting process, understand how it was built, trained, and tested. Ask vendors what data sources were used to develop the tool, how representative those datasets are, and whether the tool has been tested for disparate outcomes across demographic groups.
Employers should also ask how the vendor identifies and addresses bias, how often the model is tested, and what safeguards are in place when the tool produces potentially biased recommendations.
Once you adopt a tool you deem a fair fit, perform continuous monitoring to catch any bias concerns as applicant pools change and vendors update their underlying algorithms.
Include Human Judgment
AI recommendations should inform hiring decisions, not make them. Note that research suggests people can be influenced by biased AI recommendations. That means having a recruiter review an AI-generated recommendation isn’t always enough.
Recruiters and hiring managers should understand what information influenced the recommendation, question results that don’t align with the available evidence, and have the authority to override the system.
Provide AI Training
Hiring managers need to understand that AI recommendations can be biased and shouldn’t be accepted without question. The University of Washington study cited earlier found that implicit association testing reduced participants’ tendency to defer to biased AI recommendations by 13%.
Training can help hiring managers recognize how AI tools can fail and when to question their recommendations. It also reinforces that human oversight means actively evaluating AI outputs, not simply approving them. This gives managers the knowledge and confidence to challenge an algorithm when its recommendation doesn’t align with the evidence.
Document Your Process
Keep records of how AI tools are used throughout the hiring process, including what data they rely on, how recommendations are generated, and how those recommendations factor into hiring decisions.
Clear documentation can help employers identify problems, evaluate their processes, and respond to questions about how a hiring decision was made. Having a documented record also helps if a decision is ever challenged.
How Paycor Helps
Paycor’s Talent Acquisition software streamlines the recruitment process with AI-powered features backed by legal and compliance expertise. Recruiting features include AI-powered job descriptions, job board integrations, custom hiring workflows, and analytics.
Smart sourcing sources both candidates who are actively looking for jobs as well as passive candidates not yet in the market. Paycor WISE then calls candidates, asks your screening questions, and scores the answers automatically.
WISE allows users to control the policies, preferences, and permissions that guide AI behavior, keeping humans in the driver’s seat of the hiring process. Ready to learn how Paycor supports recruitment? Take an online tour today.