OpenAI is rolling out invisible watermarks for eligible AI-generated text in the EU to comply with the AI Act, while warning that the technology cannot determine authorship, accuracy or responsibility. The company has said that the watermark will be added to eligible text outputs over the coming weeks. ChatGPT is expected to use a system called textGrain, which changes the statistical pattern of words selected by the AI model without making the watermark visible to readers. OpenAI has detailed how the move is part of its response to the EU AI Act, which requires providers of generative AI systems to make AI-generated text identifiable in a machine-readable way. OpenAI has said it will initially make the text watermark detector available only to approved researchers and expert organisations. The company said the detector will indicate whether it finds an OpenAI watermark, but will not identify the user or reveal prompts or conversations. Should developers emulate what OpenAI is doing?
