OpenAI to Implement Watermarking for ChatGPT Output in EU

VERA Brief

AI-generated. Grounded in the article and its cited sources.

OpenAI will implement invisible watermarks for text generated by ChatGPT and Codex in the European Union. This measure is intended to address concerns about the origin and authenticity of AI-generated content.

Key facts

• OpenAI plans to introduce invisible watermarks for text generated by ChatGPT and Codex.
• The watermarking will be implemented within the European Union.
• Internal testing suggests that editing can reduce the effectiveness of the watermark.
• OpenAI is offering its API for opt-in to this watermarking feature.
• The initiative aims to address concerns about the origin and authenticity of AI-generated content.