OpenAI to watermark ChatGPT output in EU for AI Act compliance

VERA Brief

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

OpenAI is implementing invisible watermarks for ChatGPT and Codex text output, starting with users in the European Union. This is to comply with the EU's AI Act.

Key facts

• OpenAI is adding invisible, machine-readable watermarks to text generated by ChatGPT and Codex.
• This implementation will begin for users in the European Union.
• The watermarking is intended to meet the requirements of the EU's AI Act.
• The technology, named textGrain, is detectable by machines but can be harder to detect if the text is edited.
• OpenAI claims its textGrain watermarking performs comparably to or better than other methods.