Textele scrise cu ChatGPT vor putea fi depistate: OpenAI introduce un marcaj invizibil în UE. Cum vor fi identificate

OpenAI is introducing an invisible watermark called textGrain for texts generated by ChatGPT and Codex in the EU. This technology allows for automatic identification of AI-generated content, complying with the EU AI Act. The implementation will begin gradually for all eligible users, and the watermark is embedded in the text without visible alterations. A detection tool will be available only to approved researchers and organizations, not the general public, to minimize false results. While textGrain aims to improve transparency, it does not guarantee accurate detection in all cases, especially for short or heavily modified texts. The watermark indicates AI involvement but does not confirm authorship or content accuracy.