Whose Words Are These?

1. You read six vendor reports over a month. None of them states the conclusion you have reached; it came from holding all six side by side. You write it in your own words.

2. You wrote a paragraph, then used an AI assistant to fix two comma splices and cut it from 90 words to 60. The argument, the evidence and the order are unchanged.

3. A sampling method you learned on a course five years ago, now standard in your field and taught the same way everywhere. You apply it to the client's data and report what it found.

4. A published framework's four risk categories. You name the categories, then spend six pages applying them to the client's operating model with your own analysis.

5. A competitor's pricing table, taken from their public website, reproduced in your deck so the client can see the comparison you are drawing.

6. Three sentences from an academic paper that state the mechanism better than anything you would write. You want them exactly as they are.

7. An AI assistant produced the section structure for a client report — the headings and the order. You wrote every word underneath and moved two sections, but the shape is the machine's.

8. A function in code you are delivering was written by an AI assistant. You have read it, you understand it, you have tested it and it works.

9. You needed the substance of a 200-page regulatory consultation. An AI assistant summarised it. You are relying on that summary and have not read the original.

10. A slide a colleague built for a different client last year. It is excellent and it fits. You change the client name in the header.

11. A statistic you are confident about — roughly 70% of programmes fail this way. You remember reading it. Forty minutes of searching has not found the source.

12. An AI assistant produced a case study with a named company, a date and three figures. It reads perfectly. Neither the engagement nor two of the figures appear anywhere you can search.