Skip to content
inklaretaal
No. 202642

Design edition

in·klare·taal
Edition 202642 · Monday 12 October 2026 · Clarity since 2026

AI-generated text is becoming easier to spot. Hiding AI use costs trust.

What this means for you

ChatGPT text in the EU gets an invisible watermark

AI-generated text may become easier to spot. But a watermark is not conclusive proof.

OpenAI is adding an invisible watermark to text from ChatGPT and Codex. This applies to users in the EU. Over the coming weeks, eligible users will get the watermark on all subscription plans. The watermark is meant to help identify text made with AI. For you as a designer, this matters most for text that goes with visual work. Think of descriptions or campaign copy. The technology is called textGrain. The system makes small changes to the choice of words. This lets a detector spot a pattern. A detector is a program that looks for that pattern. OpenAI says the pattern remains if you copy and paste the text. So the watermark is not a visible logo, but a trace in the text. For now, only approved researchers and organisations can use the detector. The watermark makes AI-generated text easier to spot, but detection is not certain. If you change a quarter of the words, detection drops sharply, according to OpenAI. It then falls from almost 100 per cent to under 20 per cent. A detected watermark says something about the text. If there is no watermark, that does not prove that a person wrote the text. So it can help to keep track of which parts you wrote yourself and which parts were made with AI.

Source →

Background for subscribers
What this means for you
Explain how AI contributed to your design

This edition shows that AI is increasingly involved in text, images and complete campaigns. A watermark can help identify AI-generated text. It does not prove who wrote the text. A label also does not say what role AI played. At the same time, people trust a media outlet less when it hides its use of AI. So a short explanation when you deliver the work can help: which parts were made with AI, and which choices did you make yourself? That keeps it clear what AI did and what your own work is.

G-Star makes campaign film with AI from two photos

G-Star made a campaign film with techno DJ Charlotte de Witte. In the film, she climbs the Arc de Triomphe in Paris. Only two photos of the DJ were used to make it. The film was made almost entirely with AI. The creative agency says the line between coming up with an advert and making it is blurring more and more. For visual creators, this is a real example of AI being used in a whole campaign. It is not known which parts people made or changed. It may be worth watching the film and noting which parts you would want to guide yourself.

Source (in Dutch) →

Background for subscribers

This is a taster.

Subscribers read the whole Design edition: every story, each with its background in plain language.

Subscribe →

Not ready for that? Read the free letter first →

inklaretaal

AI and tech, made simple.

This is the Design edition. Subscribers get the background to every story, in plain language.
Subscribe
inklaretaal · Amsterdam · © 2026