OpenAI implements text watermarking for EU rules

OpenAI has implemented mandatory text watermarking on outputs from certain models to comply with EU AI Act provenance rules, with detection accuracy of 99.9% via a cryptographically secure watermark. The capability will roll out starting with researchers, who will be able to identify and potentially filter AI-generated content. For engineers shipping production LLM apps, API-generated text will carry deterministic token-selection constraints that could subtly alter output distributions, potentially degrading creative quality and triggering unexpected flags in users' downstream spam and plagiarism detectors.

OpenAI https://openai.com/index/eu-text-provenance OpenAI implements text watermarking for EU rules Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated. Text generated by certain OpenAI models will be detectable as AI-generated with 99.9% accuracy via a cryptographically secure watermark, and this capability will be rolled out starting with researchers, enabling them to identify and potentially filter out AI-generated content in their work. OpenAI is implementing mandatory text watermarking on its model outputs to comply with EU AI Act provenance rules, starting by opening detector access to researchers. For engineers shipping production LLM apps, this means API-generated text will contain deterministic token-selection constraints that could subtly alter output distributions, potentially degrading creative quality and triggering unexpected flags in your users' downstream spam and plagiarism detectors.