ChatGPT Is Getting Invisible Text Watermarks

Free AI writing toolkit — guide, editing prompts and AI Tells skill: https://aiwithkyle.com/mini/goblins

Join our private WhatsApp group: https://aiwithkyle.com/ai-with-kyle-group-chat

Join Freedom OS: https://aiwithkyle.com/join-freedom-os

Work with AI with Kyle: https://aiwithkyle.com/media-kit

Get the free newsletter: https://aiwithkyle.com

Subscribe: https://www.youtube.com/@iamkylebalmer?sub_confirmation=1

ChatGPT is getting invisible text watermarks. OpenAI has announced a rollout for eligible ChatGPT and Codex text in the EU, with an opt-in option for API customers worldwide.

I walk through how the pattern works, what happens when AI edits your own writing, who can access the official detectors, and what a detection result can tell you. We also look at OpenAI’s editing tests and the practical steps I would take with client work.

CHAPTERS

00:00 ChatGPT is getting text watermarks

01:14 Who gets them?

02:13 Why Claude and ChatGPT differ

03:14 How the word-choice watermark works

07:53 I wrote it. AI only edited it.

10:12 Can anyone check my writing?

13:07 Can editing weaken a watermark?

16:22 What a positive result actually proves

20:27 Did people stop caring?

22:33 What I would do with my own work

24:08 Free AI writing toolkit

SOURCES & TOOLS

OpenAI announcement: https://openai.com/index/eu-text-provenance/

OpenAI provenance help: https://help.openai.com/en/articles/8912793-provenance-signals-in-openai-generated-content

Claude text watermarking: https://www.anthropic.com/news/claude-text-watermark

Watermarks-remover repository: https://github.com/guillaumemeyer/watermarks-remover

Humanizer: https://github.com/blader/humanizer

SynthID text: https://github.com/google-deepmind/synthid-text

EARLIER VIDEOS

Claude watermarks explained: https://www.youtube.com/watch?v=rR2QW5WQ3aE

Watermark removal: https://www.youtube.com/watch?v=0tIM7yYpKuk

The EU rules: https://www.youtube.com/watch?v=eg4p7rX6dlA