ChatGPT will soon start secretly watermarking its answers — here's what that means for the text you copy and share
OpenAI is preparing to watermark ChatGPT's text — and you might never even notice it's there.
OpenAI has announced plans to begin secretly watermarking the text generated by its ChatGPT and Codex services for users located in the European Union. The invisible watermark, created using a system called textGrain, subtly alters the word choices in the AI-generated text to make it identifiable as being produced by OpenAI's model.
Currently, the watermarking feature is not accessible to the public, with only researchers and expert organizations able to utilize the accompanying AI detection tool. OpenAI claims the detector will identify whether a passage contains the watermark without revealing user information, but acknowledges that false positives and missed watermarks are possible risks.
Critics argue that the watermarking system is flawed and could potentially be circumvented, casting doubt on its effectiveness in ensuring transparency.
Written by urgent.news from TechRadar's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
