A Word of Caution on AI in Your Hiring Process

As of January 2026, the EEOC's algorithm auditing requirements are in effect. If your company uses an automated system to screen resumes, rank candidates, schedule interviews, or recommend hires, you are now expected to run annual bias audits, keep detailed impact assessments, and show measurable effort to prevent discriminatory outcomes. This is not optional paperwork. It applies whether you built the tool or bought it from a vendor.

Here is the part that catches employers off guard. Under Title VII, intent does not matter. If your AI screening tool produces a selection rate for a protected group that falls meaningfully below the rate for other groups, you are liable, even if you never meant for that to happen and never even looked at the math. The relevant benchmark is the four-fifths rule. When the selection rate for one group drops below 80 percent of the highest group's rate, the EEOC expects you to investigate immediately. Saying "the software did it" is not a defense. The agency has stated that plainly.

The enforcement picture is sobering. Algorithm based discrimination lawsuits have risen sharply since the rules took effect. In recent investigations, 74 percent of organizations could not produce proper audit documentation, and 62 percent could not show meaningful human oversight of their automated decisions. Those are the exact gaps that turn a routine complaint into an expensive legal problem.

None of this means AI has no place in your hiring. It means you need guardrails. A few practical ones:

Keep a human in the loop at every decision point. Automated rejections with no human review are the fastest route to a discrimination claim. A person should be reviewing and able to override the system, not rubber stamping it.

Know what your vendor is actually doing. If you license a screening tool, ask the provider directly how it was tested for bias, whether it has been audited, and what documentation they will give you when, not if, you are asked to produce it. If they cannot answer, that is your answer.

Keep records. Document your thresholds, your escalation triggers, and the versions of any model you use. If you cannot explain how a hiring decision was reached, you cannot defend it.

Be careful with deepfakes and identity fraud on the other side too. AI cuts both ways, and verifying that the candidate on the video call is the person who shows up to work is becoming a real part of the job.

Our view is simple. AI can help you sort and organize, but it should never be the thing that decides who gets a fire protection job at your company. The judgment, the human oversight, and the accountability still have to sit with you. Used carelessly, these tools create legal exposure that dwarfs any time they save.

If you want a hiring process that uses technology sensibly without putting your business at risk, we can help you build one.