Can you refuse to use AI at work? What to tell employers

Can you refuse to use AI at work?

You can refuse in some cases, but it's rarely a blanket right. In the US, you can ask for a religious accommodation under Title VII when required AI use conflicts with a sincerely held religious or moral belief. Outside that narrow route, refusing a tool your job requires usually counts as refusing a job duty. That line got much sharper this month. David Durman is a research administrator at the University of Kentucky who helps academics apply for grants. He won a religious accommodation that excuses him from personally using AI tools such as Copilot, ChatGPT and Claude. He's a humanist who doesn't believe in God. His argument was that doing his own thinking is part of how he develops as a moral person, and that handing it to a chatbot works against that.

The university's equal opportunity office approved the request in late September 2026. Durman had worked with a lawyer from the American Humanist Association, which says this is the first time humanism has been used to win an exemption from workplace AI. He wasn't the first person to get an AI exemption on religious grounds, though. A software engineer who is a Unitarian Universalist got a similar accommodation earlier in the year. The details matter more than the headline. Durman still goes to meetings and submits work through platforms with AI built in, because the exemption covers only his own use. The university also made the decision voluntarily, one case at a time, after looking at his duties. No court ruled on it, so it shows what one employer did, not what the law requires.

Why should a job seeker care about one administrator in Lexington? The same shift that led Durman to ask is now showing up earlier in hiring. Employers write AI tools into job descriptions, and recruiters ask 'How do you use AI?' in first-round calls. Applicant tracking systems rank your CV partly on whether it contains the AI terms in the posting. Maybe you're all in on AI, maybe you've set firm limits, or maybe you object on principle. Whichever it is, your stance now affects your keyword match before a person reads your name. The rest of this guide covers the practical side: what the law gives you, how to decode an AI requirement, what to say in interviews, and how to keep your CV competitive without overclaiming.

Does a religious exemption from AI at work cover you?

It covers you only if your objection comes from a sincerely held religious, moral or ethical belief, and only if accommodating it wouldn't cause your employer undue hardship. US federal law applies Title VII to employers with 15 or more employees. The law reads 'religion' broadly. It can include moral or ethical beliefs about right and wrong that a person holds as strongly as traditional religious views, and Durman's case relied on exactly that. Since the Supreme Court's 2023 decision in Groff v. DeJoy, an employer can't refuse just because an accommodation is a minor inconvenience. It has to show a substantial burden in the context of its business. That's a real shift in workers' favor, but employers still have plenty of room to say no.

Here's where the limits come in. A personal preference, a worry about productivity or a general distrust of tech companies won't count as a religious belief, however strongly you feel it. The job itself matters a lot too. An exemption is easy to grant when AI is a side tool, for example for a grant administrator who can write drafts by hand. It's much harder when AI is the core of the role, such as an AI support specialist or an analyst whose whole workflow runs through a model. In those cases, the honest answer is that the role probably isn't for you, and no accommodation request will change that. Outside the US the rules are different. The UK's Equality Act 2010 protects religion and philosophical belief, but no settled case has dealt with AI yet.

If you already have a job and object to an AI mandate, put your request in writing to HR or your equal opportunity office. Explain your belief and how the specific tool conflicts with it. Then offer a workable alternative, such as writing drafts yourself or having a colleague handle the steps that need AI. Keep the request narrow. Durman asked to avoid using AI himself, not to remove it from every system he touches, and that restraint probably helped. Your objection might be practical instead, such as worries about accuracy, confidentiality or client data. In that case, treat it as a policy conversation rather than a religious accommodation. Those concerns are often valid, and many teams already restrict AI tools for exactly those reasons, so you may find more support than you expect.

How should you read an AI requirement in a job description?

Put each AI mention in one of three groups: core duty, expected skill or boilerplate. A core duty reads like 'build and maintain AI-assisted workflows for the support team' or 'use generative AI to produce campaign copy at scale'. An expected skill usually shows up under preferred qualifications, for example 'comfortable with AI productivity tools such as Copilot'. Boilerplate is the line the company pastes into every posting, usually near the culture section, something like 'we embrace AI to work smarter'. Each group tells you how much AI will shape your day and how much room you have to set limits. It also shows which keywords the applicant tracking system is likely to weight most heavily when it ranks your CV against other applicants.

Where the line appears matters as much as its wording. Required qualifications usually feed knockout questions and recruiter screens. Preferred qualifications are a wish list that strong candidates often only partly meet. Pay attention to whether the ad names specific tools or just says 'AI'. A posting that lists Copilot, ChatGPT and a niche internal platform has a defined workflow in mind. A vague phrase like 'AI-forward mindset' usually means the hiring manager hasn't decided what they want. The verbs tell you a lot too. 'Build', 'own' and 'deploy' mean AI is the work itself. 'Familiarity with' or 'openness to' signals an expectation you can discuss. When in doubt, ask in the screening call: 'What does AI use look like day to day on this team?'

Here's the opinionated part: don't skip a role over boilerplate. Don't apply to a core-duty role hoping to negotiate your way out of the core duty, either. The first costs you good jobs for no reason, and the second wastes everyone's time and starts the relationship on a misunderstanding. If you want limits on AI, look at teams that handle sensitive patient, legal, financial or government data. They often restrict generative tools anyway because of confidentiality rules, so your caution looks like good judgment there. Applying across the US, the UK and India? The way AI shows up in postings varies by sector more than by country, so read each ad on its own terms. Don't assume one region is more relaxed than another.

How do you answer the 'How do you use AI?' interview question?

Describe what you actually do, the limit you set and the reason for it, in that order, and keep the whole answer under a minute. Interviewers who ask this rarely want a list of tools. They want to know whether you can tell when AI helps and when it hurts. 'I don't use AI', with nothing after it, is the weakest answer you can give. Not using AI isn't the problem. The problem is that the answer sounds like you haven't thought about it. Interviewers hear that as rigidity, even when it's a careful, principled choice you've considered for years. The fix is to show your reasoning, so the interviewer hears judgment instead of reluctance.

If you use AI with limits, try something like this: 'I use AI for first drafts of routine emails and for summarizing long documents. I don't use it for client-facing analysis or anything involving confidential data, because I need to stand behind every number. Everything it produces gets checked line by line before it goes anywhere.' If you don't use AI at all, try: 'I've made a deliberate choice to do my own drafting and analysis. I'm comfortable working on platforms that have AI built in and reviewing AI-generated work from colleagues. What I bring is careful human review, which matters more as more output is machine-written.' Both answers are honest, specific and focused on the quality of the work rather than your feelings about technology.

Don't overclaim, and don't lie. If you say you use AI daily and then freeze when a hiring manager asks for an example, that's worse than any limit you set openly. You don't have to bring up religious beliefs in an interview. In the US, Title VII also covers hiring, so an employer can't turn you down just to avoid accommodating a religious practice. Accommodation requests usually come up more naturally once you have an offer, when you can talk about the role's actual duties. For the related question of whether to disclose AI help on your CV, the same rule applies: be precise, own the final product, and never claim work you didn't do.

How do you show AI skills on a resume if you limit AI use?

List the AI-related skills you really have, in the exact words the job ad uses, and leave out anything you couldn't demonstrate in an interview. An ATS can't work out that 'careful review of automated outputs' means the same as 'AI output validation'. It matches words, and it doesn't make allowances. If the posting asks for 'experience evaluating AI-generated content' and you've spent two years fact-checking machine-drafted reports, use those words. People who limit their AI use often have exactly the skills employers are short of: reviewing outputs, writing data-privacy policies, getting human sign-off on decisions and spotting confident mistakes. Those are real AI skills, and writing them in the posting's own words lets the scanner match you fairly.

Put those terms in two places. The first is a short skills section that the parser reads cleanly. The second is at least one experience bullet that shows the result. A bare list of tool names earns you very little, a point we cover in why AI tool names alone don't impress non-tech employers. A bullet like 'Reviewed and corrected AI-drafted grant summaries before submission, catching factual errors ahead of external deadlines' is honest whether you used the tool yourself or not. Don't pad your skills list with Copilot, ChatGPT and Gemini just to match a keyword. Any decent interviewer will ask how you use them, and you'll be stuck defending a claim you never wanted to make.

How do you check your CV against the AI requirements in a job ad?

Before you send anything, check your CV against the specific posting, not just against a general idea of what employers want. Paste the job description text into HRLens's free CV analysis. For LinkedIn or Indeed ads, paste the text rather than the link, because those sites block fetching. The analysis lists the skills from the ad that your CV is missing and adjusts its suggestions to the role. It also gives an ATS compatibility score alongside your overall score out of 100. Every analysis also produces ATS-friendly rewritten versions of your CV in six templates that mirror the job's keywords, so you can see how your experience reads once the wording lines up with the ad.

Treat that output as a draft, not a final version. If a rewritten version adds an AI keyword you can't honestly claim, delete it. A small gap in your keyword match costs you less than a claim you'd have to walk back in an interview. Starting fresh, or rebuilding an old CV around this new framing? The chat-based CV builder lets you describe your experience in your own words, including your limits on AI. You can then refine each section until the wording is both accurate and easy to scan. It's also a good moment to get rid of filler phrasing that makes a CV read as machine-written, which hurts you twice if your stance is that you do your own thinking.

One last thing to check before you hit send: does your CV tell the same story you'll tell in the interview? If your CV says 'evaluated AI-generated content' and your interview answer says 'I review colleagues' AI drafts but don't use the tools myself', those fit together well. If your CV lists five AI tools and your interview answer is a principled no, the recruiter will notice the mismatch within minutes. Durman's exemption worked because his request was narrow, specific and consistent with his actual job. Apply the same rule to your CV and interview answers. Be clear about what you do and what you don't, and phrase it so both the scanner and the person reading it can tell.

Frequently asked questions