OpenAI to Add Watermarks to Some ChatGPT, Codex Text in EU
Summary
- OpenAI said it will introduce watermarks for some text generated by ChatGPT and Codex in the EU.
- The move is intended to comply with the EU AI Act's transparency requirements for generative AI.
- API customers worldwide will be able to selectively enable the watermarking feature for some supported models.
Forecast Trend Report by Period
OpenAI will introduce watermarks for some text generated by ChatGPT and Codex in the European Union as it moves to comply with transparency requirements for generative artificial intelligence under the EU AI Act.
Cointelegraph reported on October 5 that OpenAI plans to add invisible watermarks over the coming weeks to eligible text produced by ChatGPT and Codex across all subscription tiers in the EU.
OpenAI’s in-house system, called textGrain, embeds statistical signals into a model’s word choices. It does not insert hidden characters or extra spaces, and a separate detector can verify whether a watermark is present.
OpenAI said text watermarking will be enabled by default only in the EU at launch and will not become a global default setting. API customers worldwide will be able to opt in to the watermarking feature for some supported models.
JH Kim
reporter1@bloomingbit.ioHi, I'm a Bloomingbit reporter, bringing you the latest cryptocurrency news.