OpenAI will automatically watermark ChatGPT text outputs within the European Union to comply with the EU AI Act.
OpenAI announced it will automatically watermark text generated by ChatGPT within the European Union. This step helps the company comply with the EU AI Act that took effect in August, which requires machine-generated content to be marked for detection.
OpenAI announced it will automatically watermark text generated by ChatGPT within the European Union. This step helps the company comply with the EU AI Act that took effect in August, which requires machine-generated content to be marked for detection. Watermarking will remain optional outside the European Union. The proprietary system, named textGrain, inserts hidden patterns into word choices without harming output quality.
A specialized detector can find these patterns using a key. Because current standards like SynthID and the C2PA project are easily bypassed, experts believe OpenAI's system may face similar limitations. OpenAI plans to share the detector initially with a limited group of researchers and organizations. Other groups may apply for access over time through a request process.
WireUnWired turns the supplied report into a clearer brief, preserves the original publisher and author details, and adds relevant context without hiding where the information came from.
Relevant WireUnWired coverage is connected so one story can lead into the larger technology context.
Original publication: 7 October 2026 02:20
Ars TechnicaIf you're looking, here's a "good deal" for NASA and the European Space Agency
Ars TechnicaIndyCar Champion Álex Palou and Chip Ganassi Racing Utilized OpenAI to Optimize Car Setups
Ars TechnicaWikimedia Foundation Reports OpenAI Agents Attempted to Hack Wikipedia Tools and Flooded Infrastructure with Traffic