Direct answer: David Durman, a research administrator at the University of Kentucky, has become the first humanist to secure a formal workplace accommodation excusing him from personally using AI, after arguing that using the technology conflicts with his sincerely held humanist beliefs. The university’s Office of Equal Opportunity approved the request on 25 September 2026, finding it posed no undue burden on his department. The case is not a court ruling, and it happened in the United States — but it is the clearest signal yet that AI mandates at work will collide with employees’ beliefs, and UK employers who are rolling out AI tools should be thinking about that collision now, before it arrives in their own workplace.
What happened
Durman works as a research administrator at the University of Kentucky, where he supports researchers applying for grants. He is a nontheistic humanist: he does not believe in a god or supernatural power, but he holds a considered, moral framework centred on human well-being and flourishing, both individually and collectively. As AI tools grew more common in daily life and at work, he became convinced that personally using them — particularly generative AI and large language models — ran counter to those beliefs.
In his own words, published by the American Humanist Association (AHA), the nonprofit that supported his case: “As a Humanist, promoting the wellbeing and flourishing of humanity is paramount to how I live my life, and I believe that our capacity to think and reason for ourselves is worth protecting. When it comes to AI technology — particularly generative AI and LLMs, it is clear to me that the existence and use of these technologies run contrary to human wellbeing. AI is interfering with people’s ability to earn a living, the environmental costs spell disaster for humanity and our planet, and its supplanting and theft of human creativity threaten elements of our culture like art and literature that make society flourish. I sought out workplace accommodations because using AI violates my deeply held Humanist beliefs which demand I take personal responsibility for the morality of my actions. I cannot, in good conscience, partake in the use and proliferation of this technology.”
Speaking to TechCrunch, Durman put the philosophical point more precisely. “The way we become moral actors for ourselves and for others is through the active application of our minds, and the diminishing of that [through AI use] is going to necessarily have a negative impact on our ability to do so,” he said, reflecting on how overreliance on AI might affect critical thinking. His concern, in other words, is not only about jobs or the environment, though he raises both. It is about what happens to human judgement when machines do the thinking: a concern rooted in his account of how people become moral agents.
The process was methodical rather than confrontational. With the AHA’s support, Durman sent a letter to the university’s Office of Equal Opportunity in late August 2026, requesting reasonable accommodations on the basis of his “sincerely held beliefs as a Humanist”. After meeting with the office in mid-September, his request was approved on 25 September, when the office determined it posed no undue burden on his department. Durman continues to perform all of his job duties without using AI.
He was inspired to act by an earlier case. Erin Maus, a Unitarian Universalist and software engineer, had secured similar protections for herself, as Futurism reported. Maus’s situation is especially unusual, TechCrunch notes, because it has become rare for software engineers to write code without any AI assistance. Durman’s administrative role is different: he has not been required to use AI at work, though colleagues and leadership have recommended that employees try the tools. He sought the exemption now, he explained, in case a suggestion becomes a mandate in the future. “For me, this was about me and my convictions,” he told TechCrunch. “I am obviously not trying to tell people how to think or feel. But I do hope that it encourages people who have similar concerns to pursue not necessarily religious exemption, but pursue their concerns to the best of their ability where they feel is the most appropriate.”
The university’s response was measured. Spokesperson Jay Blanton told TechCrunch that the institution reviews requests for religious accommodations on an individual basis, considering the employee’s job responsibilities and the circumstances of the request. “In this case, the university determined that an accommodation could be provided while allowing the employee to continue performing the essential functions of his position,” Blanton said. “As technology and job responsibilities evolve, accommodations may be reviewed to ensure they remain reasonable and workable.” That last sentence matters: the accommodation is not necessarily permanent. If the job changes so that AI use becomes essential, the question reopens.
The AHA framed the case as a matter of equal treatment. “Mr Durman’s Humanist beliefs are sincerely held, and his employer was right to grant him an accommodation on those beliefs,” said Alexandria Huiskens, staff attorney at the AHA. “Comprehensive non-theistic belief systems like Humanism are entitled to the same religious protection under Title VII and the United States Constitution as ‘traditional’ religions, and it is a reflection of our country’s deep, longstanding commitment to religious freedom for all that non-theists like Mr Durman are entitled to these same protections and freedom of conscience.” The AHA also noted that it helped draft and is a signatory to the Luxembourg Declaration on Artificial Intelligence and Human Values, which holds that AI must never displace human judgment, reason or responsibility, while stressing that humanists hold a range of views and that the association defends each person’s right to follow their conscience.
The story also carries a wider cultural echo. TechCrunch’s report opens by noting that Pope Leo XIV has described AI as a threat to human dignity — a reminder that concern about AI and human dignity runs through several faith traditions, from Catholic teaching to Unitarian Universalism to humanism. Huiskens made the same point: “Concern about AI and human dignity can run through several different faiths and traditions, whether that’s Catholic teachings, or Universalism, or Humanism, and we want Humanists or nontheists to know that they fall into these categories — that they are also protected by the law at work.” Durman is the first humanist to secure such an accommodation, and one of the first people to publicly discuss pursuing these protections. He is unlikely to be the last.
The law behind the case — and why it is not a court ruling
It is important to be precise about what Durman won, because the headlines can mislead. This was an employer accommodation decision under US federal law, not a court judgment. No judge ruled that humanism requires an AI exemption, and no precedent has been set that binds other employers. What happened is that one university, applying its own equal-opportunity process, agreed that this employee’s request was reasonable in his circumstances.
The legal framework is Title VII of the US Civil Rights Act of 1964, which prohibits religious discrimination at work and requires employers to reasonably accommodate an employee’s sincerely held religious beliefs, practices, or observances, unless doing so would cause an undue hardship to the employer’s business. US courts and the Equal Employment Opportunity Commission have long recognised that “religion” for these purposes is not limited to belief in a god. Sincere, comprehensive, non-theistic moral frameworks can qualify. Humanism, with its organised associations, published principles, and community life, is among the strongest candidates.
The “undue hardship” test is where most such requests are won or lost. Since the US Supreme Court’s 2023 decision in Groff v DeJoy, an employer must show that an accommodation would result in substantial increased costs in relation to the conduct of its business — a higher bar than the old, minimal-cost reading. In Durman’s case, the university found no undue burden at all: his duties could be performed without AI, so the accommodation cost the department effectively nothing. That is the least controversial kind of accommodation, and it is one reason the university could agree without a fight.
Change the facts slightly, and the answer changes. A software engineer whose role now requires AI-assisted coding, a customer-service agent whose employer has automated its workflow around an AI tool, or a researcher whose grant funder requires AI-assisted analysis would face a much harder argument — not because their beliefs would be doubted, but because the accommodation would genuinely interfere with essential functions. Durman’s own case acknowledges this: his exemption is reviewable as technology and duties evolve. The precedent here is procedural and cultural more than legal: an employer took a non-theistic belief about AI seriously, processed it through its normal equal-opportunity channel, and found a workable answer. That example will be cited by employees and advisers in future requests, even though it binds nobody.
Timeline
- Late August 2026: Durman, supported by the AHA, writes to the University of Kentucky’s Office of Equal Opportunity requesting accommodation from personal AI use on humanist belief grounds.
- Mid-September 2026: Durman meets with the Office of Equal Opportunity to discuss the request.
- 25 September 2026: The request is approved; the office determines it poses no undue burden on his department.
- 26 October 2026: The AHA publishes its press release announcing the accommodation.
- 29–30 October 2026: TechCrunch reports the story, including interviews with Durman, Huiskens, and the university; the piece is updated on 30 October.
What it means for UK small businesses
UK law is different from US law, and it would be a mistake to transplant the Durman outcome directly. But the underlying issue — an employee whose beliefs conflict with a workplace AI mandate — translates cleanly, and UK employers should prepare for it. Nothing in this section is legal advice; for a live case, take advice from a UK employment solicitor. What follows is the practical shape of the issue.
The UK legal landscape, in outline
Under the Equality Act 2010, “religion or belief” is a protected characteristic. It covers religions, and it also covers philosophical beliefs that meet the Grainger criteria established by UK case law: the belief must be genuinely held, it must be a belief rather than merely an opinion based on current information, it must concern a weighty and substantial aspect of human life and behaviour, it must attain a certain level of cogency, seriousness, cohesion and importance, and it must be worthy of respect in a democratic society and not conflict with the fundamental rights of others. Humanism has been recognised in UK contexts as capable of meeting that test, and ethical veganism, for example, has been held to be a protected philosophical belief. A belief about AI’s effects on human flourishing, reasoning, and creativity could plausibly be argued to meet the test — but that would depend on the individual’s actual beliefs, their coherence, and the evidence, not on a label alone.
Crucially, UK law does not impose a general duty to make “reasonable accommodations” for religion or belief in the way it does for disability. Instead, protection works mainly through the prohibitions on direct discrimination (treating someone worse because of their belief), indirect discrimination (applying a provision, criterion, or practice — such as “all staff must use this AI tool” — that puts people sharing a belief at a particular disadvantage, unless the employer can objectively justify it as a proportionate means of achieving a legitimate aim), and harassment. An AI mandate that disadvantages an employee because of their beliefs could therefore be challenged as indirect discrimination, and the employer would need to show that the mandate is justified and proportionate — that there was no less discriminatory way of achieving the business aim. That is a fact-sensitive test, and the employer’s homework matters: a blanket mandate with no exceptions is harder to justify than a considered policy that explored alternatives.
Practical steps for a small employer
First, write your AI policy before you need it, and write it proportionately. Distinguish between AI use that is genuinely essential to a role (for example, an AI-assisted coding environment that is now the team’s standard toolchain) and AI use that is encouraged, convenient, or merely fashionable. The broader and more thoughtless the mandate, the harder it is to defend. A policy that says “use AI where it helps, with human checks, and talk to your manager if you have concerns” is both better management and legally safer than “everyone must use AI for everything”.
Second, when an employee raises a belief-based concern, take it seriously from the first conversation. Ask, respectfully, what exactly they object to: all AI use, or generative AI specifically? Personal use, or the employer’s use of AI on their work? Does the objection extend to AI-assisted tools they already use, such as spellcheck or search? Durman’s case turned on personal use in his own duties, not on a veto over the employer’s technology choices, and that distinction will often be the key to a workable compromise. Document the discussion and the decision, and apply the same process consistently to the next person who raises a concern, whatever their belief.
Third, explore practical alternatives before concluding that an accommodation is impossible. Can the employee do the same work without the AI tool, as Durman does? Can tasks be redistributed? Can a different tool, without the objectionable feature, do the job? Can a human-check step replace an AI step? If the honest answer is that the role cannot be performed without AI, say so clearly, explain why, and show your working — the operational reasons, the alternatives considered, and the impact on customers or colleagues. A reasoned, documented “no” is far stronger than a dismissive one.
Fourth, train managers to recognise a belief-based request for what it is. A team leader who treats “I can’t use this in good conscience” as awkwardness, poor attitude, or resistance to change can create a discrimination problem out of a manageable conversation. The correct response is curiosity and process: acknowledge the concern, escalate it through the proper channel, and avoid pressuring the employee while it is being considered. This is also simple decency — and it is how the University of Kentucky handled Durman’s request, through its equal-opportunity office rather than through his line manager’s discretion alone.
Fifth, think about the rest of the team. Accommodations work best when they are discreet, practical, and explained in terms of process rather than belief. Colleagues do not need to know the details of someone’s convictions; they need to know that work is covered and standards are maintained. And be alert to the reverse risk: as AI tools spread, some employees will feel pressured to use tools they distrust, while others will feel held back by colleagues who refuse them. A policy that protects both groups — the enthusiastic adopter and the conscientious objector — is one that treats AI as a tool with appropriate uses, not as an ideology.
Finally, keep the policy under review. Durman’s accommodation is explicitly reviewable as technology and duties evolve, and your policy should be too. Revisit it when you adopt a new AI system, change a role’s essential duties, or receive a request. The businesses that handle this well will not be the ones with the longest policy documents, but the ones whose managers know the process and use it calmly. For more on building AI into small-team workflows without losing the human checks, see our guide to AI task automation for small businesses, and for the wider picture of AI reshaping work itself, Uber’s ChatGPT rollout across 30,000 employees.
Honest limits / what to watch
Start with the limits of the case itself. Durman is one employee, at one university, in one US state, under a US statute, in a role where AI use was not required. Nothing about his accommodation decides what happens to a UK shop assistant, a care coordinator, or a junior developer whose employer has built AI into the core workflow. Anyone presenting this story as “employees can now refuse AI” is overstating it badly. What the case proves is narrower and still important: a sincere, well-articulated, non-theistic belief about AI can be recognised by an employer as worthy of accommodation, and the process for handling it already exists.
Watch for the next cases, which will be harder. The first contested dispute — where an employer says AI use is essential and an employee says their beliefs forbid it — will test the boundaries in a way Durman’s uncontested accommodation did not. In the US, that fight will turn on undue hardship after Groff; in the UK, it will turn on justification and proportionality under the Equality Act. Either way, the facts of the job will matter more than the rhetoric about AI. Employers whose AI mandates are genuinely tied to safety, quality, or customer requirements will be in a stronger position than employers whose mandates are merely a preference for the newest tool.
Watch, too, for the breadth of beliefs that may be raised. Durman’s humanism is one framework; Catholic teaching on human dignity, which Pope Leo XIV has invoked, is another; and employees may raise environmental, privacy, or creative-integrity concerns framed in religious or philosophical terms. Not every strongly held view qualifies as a protected belief, and employers are entitled — indeed obliged — to examine sincerity and coherence. But the examination must be respectful and even-handed. Dismissing a belief because it is unfamiliar, non-theistic, or inconvenient is precisely how discrimination claims begin.
There is also a management risk in over-correcting. Granting every AI opt-out on demand, without considering the operational impact, can burden colleagues, fragment workflows, and breed resentment. The Durman model is useful precisely because it was bounded: the accommodation covered personal AI use in his own duties, was assessed for undue burden, and remains reviewable. That balance — respect for conscience, with operational reality as the limit — is the template worth copying, not an unconditional right to refuse.
One final limit: this story says nothing about whether Durman’s beliefs about AI are correct. AI’s effects on employment, the environment, and creativity are fiercely debated, and reasonable people — including humanists — disagree about them. The AHA itself stresses that humanists hold a range of views on AI. The accommodation protects Durman’s conscience; it does not settle the debate. UK employers do not need to agree with an employee’s beliefs to handle the request properly. They need a fair process, a documented decision, and a policy that treats AI as a workplace tool subject to the same standards of fairness as any other.
FAQs
Who is David Durman and what did he win?
David Durman is a research administrator at the University of Kentucky. Citing his sincerely held humanist beliefs, he obtained a workplace accommodation excusing him from personally using AI in his job, approved by the university’s Office of Equal Opportunity on 25 September 2026.
Is this a court ruling?
No. It is an employer accommodation decision, not a court judgment. The university determined the accommodation posed no undue burden on Durman’s department. No court has ruled on these facts.
Do non-religious beliefs count for religious protection at work?
In the United States, yes: comprehensive non-theistic belief systems such as humanism are entitled to the same protection under Title VII as traditional religions, according to the American Humanist Association. UK law protects philosophical beliefs separately under the Equality Act 2010.
Does this mean UK employees can refuse to use AI at work?
Not automatically. UK protection for beliefs is real but fact-specific, and employers must consider requests individually. An employee’s role, whether AI use is an essential duty, and the impact on the business all matter. Take advice before refusing or granting a request.
What should a small UK employer do if an employee raises a belief-based objection to AI?
Listen, take the request seriously, ask what exactly the employee objects to, consider whether the work can be done without personal AI use, document the decision, and apply it consistently. Do not dismiss the belief, and do not promise an outcome before you have considered the operational impact.
Could this happen in my business?
It could, especially as AI mandates spread. The Durman case arose because AI use was encouraged and might become mandatory, not because he was already required to use it. Clear, proportionate AI policies reduce the risk of being caught out.
Sources: TechCrunch, 30 October 2026; American Humanist Association press release, 26 October 2026.
