OpenAI will start watermarking ChatGPT’s text in the EU

“OpenAI is rolling out an invisible watermark for ChatGPT and Codex text in the EU to comply with the EU AI Act’s transparency rules. The watermark, embedded in word choices, can be detected by a specialized algorithm but is not visible to readers. While the feature will be available to all EU users on all plans, developers worldwide can opt‑in via the API. The watermark does not reveal the author and has negligible impact on model performance, but it can be weakened by editing or short passages. The move follows Anthropic’s global watermarking and reflects a broader industry push to meet EU regulatory requirements. For U.S. businesses, the decision signals that similar compliance measures may soon be required, potentially increasing operational costs and influencing how AI‑generated content is marketed and audited. Expert Analysis: This regulatory compliance effort may prompt U.S. firms to adopt watermarking or other traceability solutions early, positioning them ahead of potential future U.S. legislation and enhancing consumer trust in AI‑generated content.”