AI bias in hiring happens when a recruiting tool learns from biased data and then repeats those patterns at scale. For a small business, that shows up as four main risks: biased and non-compliant decisions, legal exposure, a poor candidate experience, and missed talent. AI can genuinely help with the administrative side of hiring, but it is only as good as the data it learns from. Flawed data in, flawed hiring out.
If you are anything like me, you are always intrigued by new technology, and AI has caught my attention more than anything else in the last two decades. As in a lot of industries, AI is being sold as the future of recruiting. But as small business owners start folding it into their hiring, many are running into its real downsides. Below is what AI bias in hiring actually is, proof that it is not hypothetical, the risks worth watching, and how to use these tools without getting burned.
What AI bias in hiring actually is
An AI hiring tool does not form opinions the way a person does. It looks at the hiring data it was trained on and learns to repeat whatever patterns it finds. If those past decisions favored certain groups, even by accident, the tool learns to favor them too, and then applies that bias to every resume it touches. That is the whole problem in one line: garbage in, garbage out.
It gets worse as more job seekers use AI on their end. When applicants use AI to stuff their resumes with the exact keywords a system is looking for, every resume starts to look the same. The tool can no longer tell a genuinely qualified candidate from one who simply learned to game it, so it rewards keyword matching over real capability.
Small businesses are more exposed to this than the big players, not less. A large company can afford a team to monitor its hiring tools, test them for bias, and retrain them as the data drifts. Most remodeling shops cannot. You buy the tool, trust it, and rarely see what is happening under the hood. That gap between what the software promises and what you can actually verify is where the real danger lives, because you carry the consequences of a decision you never truly saw the tool make.
It has already happened
This is not a warning about some far-off risk. It has played out in public, at companies with far more resources than most remodelers.
- Amazon scrapped its own AI recruiting tool. In 2018, Reuters reported that Amazon abandoned an experimental hiring tool after finding it penalized resumes that included the word “women’s” and downgraded graduates of two all-women colleges. The tool taught itself that male candidates were preferable, because it trained on a decade of mostly male resumes.
- The EEOC’s first AI settlement cost a company $365,000. In 2023, the tutoring firm iTutorGroup settled with the Equal Employment Opportunity Commission after its recruiting software automatically rejected older applicants, women 55 and up and men 60 and up. It was the EEOC’s first-ever AI discrimination settlement.
- A major HR platform is now facing a nationwide lawsuit. In 2025, a federal court allowed Mobley v. Workday to proceed as a collective action on behalf of job seekers over the age of 40 who say the platform’s AI screening discriminated against them. When I wrote the first version of this article, I predicted lawsuits were coming. They are now here.
The real risks of AI in recruiting
Strip away the overlap and the risks fall into six clear buckets. Any one of them can cost you a great hire or land you in legal trouble.
- Garbage in, garbage out. As resumes get more optimized, the tool gets worse at spotting real talent and better at rewarding whoever gamed it best.
- Built-in bias that can break the law. AI inherits the biases in its training data and can amplify them, disadvantaging protected groups and exposing you to discrimination claims under laws like Title VII, the ADEA, and the ADA.
- A poor candidate experience that hurts your brand. When people feel screened out by a faceless system, they feel undervalued, and word gets around. That makes it harder to attract good people the next time you hire.
- Overlooked talent from keyword dependence. Strong candidates who do not use the exact phrasing the system expects get filtered out. AI cannot see transferable skills, potential, or culture fit the way a person can.
- Data privacy exposure. These tools process a lot of personal information. Mishandling it can lead to breaches, legal headaches, and reputational damage.
- No read on the human stuff. AI cannot measure passion, adaptability, or whether someone shares your company’s values, and in a small business those qualities often matter most.
How to use AI in hiring safely
None of this means you have to swear off AI. It means you use it with your eyes open. A few ground rules keep it an asset instead of a liability:
- Keep a human in the loop. Use AI to sort and organize, not to make the final call. A person should review candidates and own every hiring decision.
- Use it for admin, not judgment. Scheduling, note-taking, and drafting job posts are safe. Ranking and rejecting people is where the risk lives.
- Ask hard questions before you buy a tool. Find out what data it was trained on, whether it has been tested for bias, and how often it is audited. If a vendor cannot answer, that is your answer.
- Know the rules that apply to you. There is no single federal AI hiring law, and the EEOC scaled back its AI guidance in 2025, but a growing patchwork of state and local rules now applies, including New York City’s bias-audit requirement, Illinois’ 2026 AI employment rules, and Colorado’s AI Act. The bigger point: the anti-discrimination laws that already exist apply to an AI decision exactly as they would to a human one.
- Protect candidate data. Collect only what you need, store it securely, and know how the tool uses it.
- Vet your recruiting partner. If you outsource recruiting, ask how they use AI and who is accountable for the results. As the Workday case shows, using a vendor’s tool does not move the liability off your desk.
The bottom line on AI bias in hiring
AI in recruiting is not the silver bullet it is often made out to be. It can help with certain administrative tasks, but it introduces real risks to both your hiring and your legal standing. Use it cautiously, and always balance the technology with human judgment. If you use an outside recruiting service, be especially wary of anyone rushing to jump on the AI wagon. You could be taking on liability with very little visibility into how they train and use their tools, and the litigation over exactly this is already underway.
Call me old school, but until these tools mature and sensible regulations catch up, nothing replaces the insight and intuition of an experienced recruiter who understands the laws, the compliance landscape, recruiting strategy, and the real needs of a business. That human judgment matters even more as the skilled labor shortage makes every hire count.
AI bias in hiring: frequently asked questions
Is AI biased in hiring?
It can be. AI hiring tools learn from past hiring data, so if that data reflects biased decisions, the tool tends to repeat and even amplify them. Real cases bear this out: Amazon scrapped a recruiting tool that penalized women’s resumes, and the EEOC’s first AI discrimination settlement involved software that auto-rejected older applicants. Bias is not guaranteed, but it is common enough that every AI tool needs testing and human oversight.
What are the risks of AI in recruiting?
The main risks are biased or discriminatory decisions, legal and compliance exposure, a poor candidate experience that damages your brand, qualified people being filtered out over keywords, data privacy issues, and the tool’s blindness to soft skills and culture fit. For a small business, any one of these can mean a bad hire or a legal problem.
Is AI recruiting legal?
Yes, using AI in recruiting is legal, but it does not lower your responsibility. Existing anti-discrimination laws, including Title VII, the ADEA, and the ADA, apply to an AI-driven decision just as they do to a human one, and a growing set of state and local laws adds bias-audit and disclosure requirements. If an AI tool produces a discriminatory outcome, the employer can be held liable, even when a vendor built the tool.
Article written by Erin Longmoon – CEO & Founder of Zephyr Connects
