A consultant at a management consultancy uses Claude in several ways during one week. Her firm's AI-use policy asks staff to tell the recipient about AI involvement whenever the recipient would reasonably rely on the work as the consultant's own professional judgement. Which of these situations most clearly calls for her to tell the recipient?
- AClaude corrected the grammar and spelling in an internal email she wrote to arrange a team meeting.
- BClaude turned her handwritten workshop notes into a tidy table for her own follow-up, which no one else will see.
- CClaude suggested ten headline options for a client newsletter, and she wrote her own headline using none of them.
- DClaude drafted most of the risk assessment a client is paying for as her expert opinion, which she lightly edited. Correct
Why A is wrong: Any AI use can feel as though it needs flagging. But correcting grammar in a meeting request does not change the substance, and no recipient relies on it as her professional judgement.
Why B is wrong: The table is AI-assisted, which may prompt a careful colleague to disclose it. There is no recipient here at all, so the policy's test of reliance by a recipient does not apply.
Why C is wrong: Claude was involved in the process, so disclosure may seem prudent. However, none of Claude's suggestions reached the client, and the published headline is entirely her own work.
Why D is correct: The client is paying for her professional judgement and will rely on the assessment as such, yet most of it came from Claude with light editing. This is the situation the policy describes, so she should tell the client.