OpenAI's text watermarking is easy to erase: swapping 25% of words drops detection to 17%
Our approach to EU text provenance rules
OpenAI is rolling out textGrain, an invisible watermark for ChatGPT and Codex text in the EU, and opening detector access to approved researchers. The technology performs well on long passages but struggles with short, constrained text like math, and editing can weaken the signal dramatically. OpenAI says watermarking doesn't measure human contribution, establish ownership, or verify accuracy, and it won't be a global default at launch.
In an evaluation of 400-token passages, replacing 10% of words with synonyms reduced detection from about 92% to 66%. Replacing 25% of words reduced it to 17%.
- socketcluster
This approach feels wrong. For code, it's obvious now that Claude is adding watermarks through comments because they are way too long. I keep asking Claude to remove and reduce its comments.
Also, I don't think it's useful because AI involvement in the work is a spectrum, not a binary true or false. If I ask Claude to go over my code and fix up typos and clean up and it starts adding comments with hidden watermarks everywhere it makes the tiniest adjustment, it's essentially appropriating my code. Its contribution may be trivial and superficial.
When I use AI on my own projects, the code it generates basically looks like auto-complete to me; exactly what I would have written by hand. So I don't see why it needs to be marked as AI. It's just saving me time. It's very much my own work.
- LudwigNagasena
It’s obvious that such boneheaded methods don’t work. So what’s next? First, we need to deal with basic word substitution, which seems theoretically feasible. Then we need to deal with encodings such as Caesar cipher, replacing spaces with zero-width spaces, Base64, etc. Your account will be flagged and reported for outputting obfuscated text. What’s next after that? Ooops, you output too many vim commands instead of outputting text directly, your account is flagged and reported to Europol. You think you can bypass that with Deepseek? No, it will be banned alongside VPN.
- Aerroon
Could this technique be theoretically used to track users themselves? It would be quite ironic if the EU forced tech companies to implement extra tracking, wouldn't it?
- mgax
This is such a waste of time.
If someone wants to bypass this it will be rather simple. Just change the words. If someone wants to avoid fingerprinting they will.
Can’t we just focus on building rather than spending brainpower on these ridiculous sidequests
- m-hodges
> Starting today, API customers globally will be able to opt in to text watermarking for select models. Text watermarking will remain off by default in the API.
> Over the coming weeks, we will add an invisible watermark to eligible ChatGPT and Codex text output in the European Union.