OpenAI plans to watermark text produced by ChatGPT and Codex for users in the European Union, according to TechCrunch AI. The company says the step is intended to comply with the EU AI Act. The watermarks are described as invisible marks embedded in generated text, though OpenAI acknowledges that editing the output can make them harder to detect.
Why it matters
Watermarking AI-generated text is one approach to making machine-produced content identifiable, a concern addressed by the EU AI Act. OpenAI’s own caveat about editing weakening detection highlights the practical limits of such measures.
Who should care
EU-based users of ChatGPT and Codex, as well as organizations tracking AI transparency and regulatory compliance requirements, should note this change.