← Founder Notes
Archive

The watermark just arrived with a warning. openai is adding an invisible watermark to chatgpt and…

Yethikrishna ROriginal on Threads

the watermark just arrived with a warning. openai is adding an invisible watermark to chatgpt and codex text in the eu, with api opt-in, while its own tests admit that editing can weaken it.

provenance is a first draft.

Context

OpenAI's post of October 5, Our approach to EU text provenance rules, says it is adding watermarks to text in response to the EU AI Act. Search Engine Journal says eligible ChatGPT and Codex text in the EU gets an invisible watermark, API users can opt in on select models and detector access starts with approved researchers.

BleepingComputer says the watermark is invisible when you read or copy the text.

How it compares

The EU rollout, the API opt-in and the invisible watermark match OpenAI and Search Engine Journal.

OpenAI's own tests show synonym edits and short passages make the mark harder to detect, which matches the note's 'editing can weaken it'.

'Provenance is a first draft' is the author's opinion.

Related work

Watch next

  • Read OpenAI's post for the detection limits it lists.

Sources

  1. OpenAI, Oct 5, 2026openai.com
  2. BleepingComputer, Oct 5, 2026bleepingcomputer.com
  3. Search Engine Journal, Oct 6, 2026searchenginejournal.com
  4. TechRepublic, Oct 6, 2026techrepublic.com

Provenance

The note above is reproduced unedited from the original post, first published on Threads on 10 October 2026 at 05:08 IST. Sources are the papers and datasets the note draws on.

View the original post
Embed this note
<iframe src="https://founder.myndlabs.tech/notes/embed/the-watermark-just-arrived-with-a-warning-openai-DeStQ6bDK3d" width="480" height="420" style="border:0;max-width:100%" loading="lazy" title="The watermark just arrived with a warning. openai is adding an invisible watermark to chatgpt and…"></iframe>

More notes