本文目前仅提供英文版本。
Can I Refuse AI at Work for Professional or Ethical Reasons? A Practical Guide
What if you believe the AI your employer is introducing is inaccurate, biased, or unethical? What if using it would conflict with your professional obligations? A plain-English guide to when refusal is justified — and how to do it effectively.
Key Takeaways
Refusing to use an AI tool your employer has introduced is a significant step with potential disciplinary consequences — it should be a last resort after documented internal escalation has failed.
Professional obligation conflicts are the strongest grounds for refusal: if a regulated professional (lawyer, accountant, doctor, financial adviser) believes using an AI tool would breach their professional obligations, that is a genuine basis for raising concerns and potentially for refusal.
Accuracy and safety concerns — particularly in regulated contexts — can justify documented refusal to rely on AI outputs without adequate verification, even if you cannot refuse to use the tool entirely.
Ethical concerns about an AI tool (bias, unfair outcomes, privacy violations) are legitimate grounds for formal internal complaint but generally are not standalone grounds for disciplinary-risk refusal, unless the use would itself be unlawful.
Document everything: your specific concern, who you raised it with, when, their response. This documentation protects you if the matter escalates and demonstrates good faith engagement rather than simple non-compliance.
"仅供参考。本文不构成法律、监管、财务或专业建议。如需具体指导，请咨询合格专家。"
The distinction between concern and refusal
The most important practical point: there is a significant difference between raising concerns about an AI tool and refusing to use it. Raising concerns is almost always the right first step and carries minimal risk. Outright refusal — particularly without prior internal escalation — carries disciplinary risk that could be avoided by a more graduated approach. In most employment contexts, the right path is: raise your concern formally in writing, propose a specific remedy (additional verification requirements, limitation of AI use to lower-stakes tasks, independent accuracy assessment), and escalate internally if the concern is not adequately addressed. Refusal should be reserved for situations where the tool's use would be directly unlawful or where internal escalation has genuinely failed.
When professional obligation provides a basis
For regulated professionals — solicitors, barristers, accountants, financial advisers, registered nurses, registered medical practitioners — professional conduct rules create duties that can provide genuine grounds for declining to use AI in specific ways. A solicitor whose professional indemnity insurer has not approved a specific AI tool for use with client data may have professional grounds for declining to use it. A financial adviser whose AI-generated advice tool has not been authorised under their AFS licence may have regulatory grounds for declining to rely on its outputs without additional verification.
The professional obligation ground works best when you can point to a specific rule, standard, or insurer requirement — not a general sense that the AI might not be trustworthy. Professional bodies in most regulated occupations have issued or are developing AI-specific guidance that can provide the specific basis you need.