OpenAI will begin adding invisible watermarks to text generated by ChatGPT and Codex in the European Union, as it moves to comply with the EU AI Act’s transparency rules.
The watermark is expected to roll out over the coming weeks for eligible ChatGPT and Codex users on all plans in the EU. Developers using OpenAI’s API globally can enable the feature on select models, although it will be turned off by default. OpenAI said it will not make text watermarking a global default at launch.
Rather than appearing as a visible symbol, the watermark subtly influences word choices to create a pattern that detection systems can identify. The pattern remains in the text when it is copied and pasted, and does not reveal the user’s identity, according to the company.
OpenAI has also published a technical report on the system, called textGrain. The company said its tests showed no meaningful impact on model performance, but found that editing, short passages, mathematical answers and translated text can make detection more difficult. Replacing 10% of words with synonyms reduced detection in one test from about 92% to 66%.
OpenAI said the watermark will initially be evaluated by approved researchers and expert organisations. It also cautioned that the absence of a watermark does not prove human authorship, while its presence cannot show how much human judgment or editing went into a passage.
Source: TechCrunch
