OpenAI says it will add invisible text watermarks to eligible ChatGPT and Codex outputs in the EU as part of its response to the EU AI Act’s provenance requirements.
OpenAI says it will add invisible text watermarks to eligible ChatGPT and Codex outputs in the EU as part of its response to the EU AI Act’s provenance requirements.
Its “textGrain” system embeds a statistical signal into word choices, while API customers worldwide can opt in to watermarking on supported models. OpenAI also warns that detection becomes less reliable for short or edited text and that a watermark does not establish authorship, ownership, or factual accuracy.