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.
