OpenAI unveils textGrain: A new watermark for AI-written text

OpenAI has revealed textGrain, a new system for watermarking text. The name of the author behind the original coverage is Pasquale Pillitteri.

The idea of a text watermark is broadly that text from an AI can get a hidden mark, so you can later see that it comes from an AI. How textGrain does this specifically, and how well it works, is not clear from the material we have received.

What does it mean for you personally? Nothing for now. You do not need to change the way you use AI, and you should not count on being able to detect AI text with certainty yet.

What does it mean for your company? If you use AI for texts, watermarks could eventually become relevant, for example in relation to customers, partners or internal guidelines. But that is the future, not a task for today.

Rule of thumb: Wait calmly until there is concrete information about when textGrain will be put into use and what it means for the tools you already use. Only then does it make sense to check your routines and any AI policy.

What it means for you

If you need to know whether a text was written by AI, a watermark may eventually make that easier, but it is not something you can use yet. You do not need to change how you use AI for emails, reports or anything else in everyday life. If you work at a company that uses AI for texts, it is a good idea to keep an eye on news about textGrain and wait to adjust your routines or AI policy until more concrete information arrives.