University of Kentucky Grants Humanist an AI Opt-Out | Zeniteq
Loading
Loading article
Zeniteq is now available on mobile for iPhone, iPad and Apple Watch.Download nowLearn more
University of Kentucky Grants Humanist an AI Opt-Out
A research administrator’s accommodation shows how non-theistic beliefs can intersect with workplace AI policies, while leaving employers to assess each request individually.
The University of Kentucky recently granted a research administrator a religious accommodation to opt out of using AI, citing his sincerely held Humanist beliefs. Under Title VII of the Civil Rights Act, employers must provide reasonable accommodations for non-theistic moral or ethical convictions, provided they do not cause substantial operational hardship. Because the employee could still perform his essential duties without AI, the university avoided a legal conflict. This case highlights that employers should distinguish between required job outcomes and specific tool mandates, as individual assessments are necessary when AI policies intersect with protected beliefs.
The University of Kentucky has granted research administrator David Durman a workplace accommodation excusing him from personally using AI, following a request based on his Humanist beliefs. The American Humanist Association, which supported his request, announced the accommodation on October 26, 2026. The university approved it on September 25.
One detail matters especially for employers considering AI mandates: Durman had not yet been required to use AI. He sought protection before encouragement to try the tools could become a requirement. That makes this an early example of addressing a potential conflict through an accommodation process, without waiting for a disciplinary dispute.
What the University Approved
Durman supports researchers applying for grants, according to TechCrunch. He approached the American Humanist Association, or AHA, because he believed using AI conflicted with his ethical commitments.
Late August 2026: Durman sent a letter to the university’s Office of Equal Opportunity requesting reasonable accommodations.
Mid-September: He met with the office to discuss his request.
September 25: The office approved the accommodation, finding that it posed no undue burden on his department.
October 26: The AHA publicly announced the outcome.
The AHA says Durman continues to perform all his job duties without using AI.
Durman connects his objection to Humanism’s emphasis on human wellbeing, independent reasoning and personal moral responsibility. In the announcement, he raises concerns about AI’s effects on employment, the environment and human creativity, particularly in relation to generative AI and large language models.
Those concerns explain his convictions; the accommodation is not an adjudication of the underlying claims about AI’s effects. An employer can evaluate a belief-based request without settling arguments about the technology’s overall social value.
The AHA also says Humanists hold different views about how to act on AI-related concerns. Durman’s position should therefore be understood as his own application of his beliefs, not a universal Humanist prohibition on AI use.
Why Non-Theistic Beliefs Can Qualify Under Title VII
Title VII of the Civil Rights Act of 1964 protects employees against religious discrimination and requires covered employers to provide reasonable religious accommodations unless doing so would create undue hardship. It generally covers employers with at least 15 employees, according to the Equal Employment Opportunity Commission’s workplace guidance.
Belief in a god is not a prerequisite. The EEOC explains that protected religious beliefs can include non-theistic:
“moral or ethical beliefs as to what is right and wrong which are sincerely held with the strength of traditional religious views.”
That definition helps explain how a Humanist employee can seek a religious accommodation without claiming supernatural beliefs or joining a conventional religious institution.
The same guidance draws an important boundary. Strong conviction alone does not make a belief religious for Title VII purposes. Social, political and economic philosophies, along with personal preferences, are outside that category.
For AI policies, the distinction is consequential. An employee’s dislike of a chatbot, concern about subscription costs or disagreement with a company’s automation strategy does not automatically create an entitlement to a religious accommodation. An objection rooted in a qualifying, sincerely held belief system calls for a different analysis.
The subject of an objection is not enough to resolve that distinction. Concerns about jobs or creativity might form part of one employee’s religious or ethical commitments and another employee’s policy preferences. The EEOC guidance explains that the same practice can have religious significance for one person and purely secular significance for someone else.
Nor must every member of a belief community agree. The EEOC’s regulations say that a religious group’s failure to endorse a belief does not determine whether it is religious for the individual employee.
My view is that this is the most useful legal distinction in the case: the inquiry concerns the employee’s beliefs and their relationship to the work requirement. A label such as “Humanist,” “religious” or “anti-AI” cannot do all that work by itself.
The University’s Decision Was Tied to Durman’s Duties
University spokesperson Jay Blanton told TechCrunch that Kentucky reviews religious accommodation requests individually, taking account of job responsibilities and the circumstances of each request. In Durman’s case, it determined that he could continue performing the essential functions of his position with the accommodation.
Blanton also said:
“As technology and job responsibilities evolve, accommodations may be reviewed to ensure they remain reasonable and workable.”
That qualification makes the arrangement dependent on actual working conditions. It is not an unconditional promise that every future version of Durman’s role will remain compatible with the same accommodation.
A future review would not, by itself, mean the accommodation should end. It would mean reconsidering the conflict, available alternatives and operational consequences under the circumstances then in place.
Durman’s position also differs from a situation where an employee has already refused a mandatory AI-dependent assignment. TechCrunch reported that colleagues and leadership had recommended trying AI tools, but he had not been required to use them. The published outcome therefore offers limited evidence about how the university would handle a role in which using a particular AI system was already integral to the work.
The Hardship Standard Changed After Groff
An employer’s accommodation obligation has limits, but older summaries can understate it.
The EEOC’s religious-discrimination compliance manual was issued on January 15, 2021. A prominent notice now explains that the Supreme Court’s 2023 decision in Groff v. DeJoy supersedes contrary material on the page.
According to that EEOC notice, the Supreme Court held that:
“undue hardship is shown when a burden is substantial in the overall context of an employer’s business”
The Court rejected the idea that a cost exceeding a minimal amount is enough. The assessment must consider the accommodation’s practical effects alongside factors including the employer’s nature, size and operating costs.
This matters because a non-AI workflow could involve some inconvenience or additional work without necessarily reaching the legal threshold for undue hardship. Equally, a belief-based request does not erase substantial operational burdens.
The relevant question is what the proposed accommodation would actually require. Would the employee still deliver the required work? Would an alternative process affect deadlines, staffing or costs? How significant would those effects be in that employer’s operations?
The EEOC’s regulations also address a likely concern for organizations introducing AI widely: assuming that many more employees might request the same accommodation is not evidence of undue hardship. They call for considering available alternatives and whether those alternatives would actually create hardship.
For an employer, a general prediction that exemptions could undermine an AI rollout is weaker than a documented assessment of the specific work involved. That is a practical implication of the published standards, not a finding about any dispute at Kentucky.
What Employers Should Define Before Mandating AI
The distinction between a job’s required output and its preferred tools is central to this case.
Kentucky’s account says Durman can perform his essential functions with the accommodation. That supports a narrow but useful inference: in his current circumstances, personally using AI is separable from successfully doing the job.
I’d make that distinction explicit before drafting an AI-use mandate. Requiring an accurate grant document by a deadline describes an outcome. Requiring an employee to generate its first draft with an AI assistant specifies a method. Those requirements can create different accommodation questions, even when management expects the same final deliverable.
A practical review could address the following points:
Review Point
Why It Matters
Identify the exact requirement
A named tool, a particular task and a general expectation to “use AI” create different conflicts.
Establish the employee’s objection
The employer needs to understand which activity conflicts with the employee’s beliefs.
Consider workable alternatives
A non-AI process may preserve the required result, depending on the role.
Assess concrete operational effects
Time, cost, staffing and other consequences need to be evaluated in the employer’s context.
Record the scope and review conditions
Both parties need clarity about what is covered and what changes could warrant another assessment.
These are practical applications of the case and EEOC framework, not a claim that Kentucky used this exact checklist.
“Personal AI Use” Needs a Clear Boundary
The published accounts describe an accommodation from personal AI use but do not provide detailed terms defining every covered tool or activity. They do not resolve how the arrangement treats embedded AI features, AI-generated material supplied by colleagues or software with optional AI functions.
Those distinctions could matter in another workplace. Asking an employee to enter prompts into a generative AI service is different from asking them to process a document that someone else produced with AI. Whether either activity creates a belief conflict depends on the individual request.
Employers should avoid assuming that “AI” identifies one clearly bounded activity. A useful policy would name the relevant systems and tasks, explain what use is mandatory, and distinguish required functions from optional features.
The same precision helps employees. A request explaining the conflicting activity and a workable alternative gives an employer more to evaluate than a general objection to AI.
The Kentucky case’s practical contribution is a documented example of accommodation while the employee continues performing his duties. For employers watching the AI industry’s push toward wider adoption, the useful lesson is to assess the work and the individual conflict before assuming that a uniform tool mandate is necessary.
Frequently Asked Questions
3 questions
1
Does This Give Every Humanist Employee an AI Exemption?
No. Kentucky approved an individual request. The employee’s beliefs, the work requirement, available accommodations and potential hardship all matter.
2
Can Nonreligious Employees Request Religious Accommodations?
Yes. EEOC guidance includes qualifying non-theistic moral or ethical beliefs. A personal preference or strongly held policy opinion alone is insufficient.
3
Can Kentucky Revisit Durman’s Accommodation?
Yes. The university told TechCrunch that accommodations may be reviewed as technology and responsibilities change. A review would assess whether the arrangement remains reasonable and workable; it would not automatically end it.
