Your ad reads fine, but you can't tell whether it quietly tells some people not to apply. An AI tool can't settle that for you. It can only help you spot phrases to look at.
Why a draft can go wrong
A UK government guide on AI in recruitment says there is "a risk of unfair bias or discrimination against applicants".1 A tool writes from patterns in text it has seen. It doesn't know your role, your law or your workplace. So a draft can carry phrases that sound normal and still put people off.
That guide also says it is not "legal assurance or legal advice".1 The same caution fits AI chat answers. For a legal answer, ask the official body or a qualified adviser.
Step 1: Find phrases that describe a person
Read the ad and mark every phrase that describes the kind of person you want, instead of the work to be done. Look for words about age, sex, family, background, health or looks. Examples are "young and energetic", "a digital native" and "must be able-bodied". Each one needs a second look.
Then ask whether the job truly needs it. A role that needs lifting should say what must be lifted. A role that needs a licence should name the licence. Say what the work is, and leave out what kind of person you imagine doing it.
Step 2: Check against the official rules
In the United States, the EEOC says it's illegal to publish a job ad that "discourages someone from applying for a job".2 That covers race, colour, religion, sex, national origin, age (40 or older), disability and genetic information.2
It gives an example. An ad seeking "females" or "recent college graduates" "may discourage men and people over 40 from applying and may violate the law".2
In Australia, an employer "can't take adverse action against an employee or a prospective employee for discriminatory reasons".3 That's the Fair Work Ombudsman's wording. It also says people are "protected from adverse action because of protected attributes".3 The page lists those attributes, including age, sex, pregnancy, religion and physical or mental disability.3
Those are two countries' rules. They aren't the same, and neither one is a full list of what applies to you. States, provinces and other countries add their own. A separate page lists where each official body publishes its rules.
Step 3: Check the requirements too
Bias hides in requirements as well as in descriptions. Look at each "must have" and ask what it screens out.
- Years of experience. Does the job really need that many, or would a skill test show the same thing?
- Qualifications. Is the degree needed, or is it a habit?
- Physical demands. Do they describe real tasks of the job?
- Tests. The EEOC says that if you require a test, "the test must be necessary and related to the job".2
If a requirement is there only because older ads had it, cut it or move it to "nice to have".
Step 4: Use the AI for a second opinion only
You can paste the ad into a chat and ask it to flag wording that might discourage some groups from applying. Treat the answer as a list of things to look at. It may miss problems, and it may flag things that are fine.
When it suggests a change, check the change against the official source yourself. If you're still unsure, a person who knows the law where you are should look at it before the ad goes out.
If an applicant raises a concern
Take it seriously and look at the ad again. Official bodies handle complaints, and you can read what they say about your duties on their own sites. Don't rely on a chat tool to tell you whether the ad was lawful.
Reference list
- Responsible AI in recruitment, Department for Science, Innovation and Technology (UK)
- Prohibited Employment Policies/Practices, U.S. Equal Employment Opportunity Commission
- Protection from discrimination at work, Fair Work Ombudsman (Australia)