At talks on content ethics, one demand comes up almost every time: companies should disclose whether their content came from a person or from AI. It sounds obviously right. Readers deserve to know who or what wrote the words in front of them.

Then I look at who is making the demand. Consultancies, agencies, thought leaders, the “responsible” companies that put the case forward at these same talks. I check their own websites, their concept papers, their newsletters. Nowhere does any of them say how their own content came about. Whether a piece was AI-generated, AI-edited or written by a person from start to finish stays unstated, every time.

Some hint that AI does not always deliver the quality their work requires. That hint tells me nothing about what AI actually does for them: research, drafts, phrasing, structuring, summarising studies. Any of the five could be happening behind a piece of their writing, and they disclose none of it.

The demand for transparency is easy to make about someone else’s content. Applied to the person making it, the demand runs into the nuances and grey areas they would rather not show. I know companies that loudly demand transparency but don’t like to talk about how they themselves work.

Experts who demand a standard they don’t meet themselves should lose credibility for it, yet they don’t. Some of these same advisers recommend specific wording for disclosure labels, phrases such as “created with AI assistance” or “this content was partially auto-generated”. The wording is plausible enough that a reader would trust it on sight. Nobody knows whether the advisers use such a label on their own output, because they never say, even as they explain why everyone else should.

This looks less like calculated hypocrisy or a deliberate lie than a blind spot, one that forms when someone writes rules for others and does not count themselves among those the rules apply to. The adviser analyses the system and recommends changes to it, then stops, without asking where their own output fits inside it.

Every document today draws some suspicion of machine authorship, whole or partial: blog posts, news articles and the writing of the AI advisers themselves. Writing about AI’s role in how content gets made while staying silent about AI’s role in one’s own does not meet the standard these advisers recommend to everyone else.

Transparency has to start with the person asking for it. Among the advisers whose websites I checked, none states what AI did to their own text, so the standard they recommend still runs without a single working example.