AI‑produced text or media shown to users as if a human created it, without a clear label.
01What it is and how it works
An AI model generates a piece of content—blog post, product description, video script, etc. The publisher then publishes that output without indicating that a machine created it. The user sees the content and assumes a human author, because no disclaimer, attribution tag, or schema markup says otherwise. The lack of disclosure can affect trust, especially when the content influences purchase decisions or brand perception.
It is AI content that looks human but isn’t labeled.
02What to do about it
Take these steps this week to eliminate undisclosed generation from your brand’s output:
- Add a clear disclosure line (e.g., "This article was generated by AI") at the top or bottom of every AI‑written asset.
- Use schema.org’s
authorproperty to point to a human editor or to anOrganizationthat oversaw the AI output. - Create an internal checklist that flags any content produced by tools like ChatGPT, Claude, or Gemini before it goes live.
- Train your copy team on the brand’s disclosure policy and keep the policy in a shared document.
03How it is measured or noticed
Our platform looks for three signals: (1) missing AI disclosure text, (2) absence of author or creator schema that references a human, and (3) language patterns that match known AI model outputs (e.g., repetitive phrasing, lack of personal anecdotes). When any of these signals appear, the content is flagged as potential undisclosed generation. You can also run a manual audit by searching for phrases like "generated by" or checking the page source for data-ai-generated attributes.
04Common mistakes
- Leaving the disclosure in a tiny footer that users can’t see without scrolling.
- Using vague language such as "assisted by AI" when the AI wrote the entire piece.
- Adding a disclosure only in the CMS notes but not on the published page.
- Relying on a single AI model’s watermark and assuming it covers all generated assets.
05Limits
Undisclosed Generation does not apply to content that is purely human‑written, even if a human used AI for research. It also does not cover images that are AI‑generated when the image file includes proper EXIF metadata indicating the source. The term is often confused with "AI‑assisted" content, where a human author edits or curates AI output; the key difference is whether the final piece is presented as human‑authored without any mention of the AI role.
06Worked example
"We used ChatGPT to draft the product description, then added a line at the end: ‘This description was generated with the help of AI.’ The page also includes <script type="application/ld+json">{\"@type\":\"Product\",\"author\":{\"@type\":\"Person\",\"name\":\"Jane Doe\"}}</script> so search engines see a human overseer."07Frequently asked questions
How does Undisclosed Generation differ from disclosed AI content?
It differs because Undisclosed Generation is AI‑produced material presented as human‑created without any clear label, while disclosed AI content includes an explicit statement that it was generated by artificial intelligence. The former hides the origin, which can mislead users, whereas the latter is transparent about the use of AI.
Should we disclose every piece of AI‑generated content, or are there exceptions?
Yes, you should disclose any content that was directly created by an AI model and is shown to users as if a human wrote it. The only exception is when the AI was used solely for research or inspiration and the final output is entirely human‑written.
Who is responsible for labeling AI‑generated content on our brand’s website?
Usually the content owner or the publishing team is responsible for adding the disclosure. If a third‑party agency creates the material, the contract should require them to include the label before delivery.
What are the consequences if we fail to label AI‑generated content?
It can damage trust with your audience and may lead to regulatory scrutiny in jurisdictions that require transparency. You’ll also notice higher bounce rates or negative feedback when users discover the omission.
How can we detect Undisclosed Generation in existing assets?
Our platform scans for three signals: missing AI disclosure text, lack of a human‑referencing author or creator schema, and language patterns that match known AI models. When those signals appear together, the content is flagged for review.
Does using AI for research but writing the final copy myself count as Undisclosed Generation?
No, it does not count because the final text is entirely human‑written. The rule only applies when the AI‑generated output itself is presented to users without a disclosure.
08Asked out loud
The same term in the words somebody uses speaking to an assistant rather than typing into a box — written from the situation, which is why each one carries the situation it came from.
Yes, you should add a clear disclosure before sharing the description. Presenting AI‑written copy as human‑created can erode trust and may violate transparency guidelines.
It depends on whether the paragraph was directly generated by an AI model and shown without attribution. If it was, you need to add a disclosure; if a human edited it substantially, it may not require one.
Usually you should check the page’s metadata for an author or creator schema and look for any AI disclosure text. If neither is present, treat it as potential Undisclosed Generation and add a label before the deadline.