OpenAI to Implement Watermarking for ChatGPT Text in Compliance with EU AI Act

OpenAI will start watermarking text generated by ChatGPT in the European Union to comply with the EU AI Act's transparency rules that took effect on August 2. This follows Anthropic's announcement two months prior regarding watermarking for its Claude model. OpenAI's watermarking aims to indicate whether content was generated or processed by its systems, although it will not specify the extent of human input involved. The company reported a 92% detection rate for AI-generated content before editing, which drops to 66% after replacing 10% of words with synonyms. OpenAI plans to provide initial access to its detection tools only to approved researchers and expert organizations to evaluate their reliability and responsible use. This move aligns with OpenAI's ongoing shift towards ethical AI development, as highlighted in previous coverage. The watermarking initiative may influence how developers and researchers approach AI-generated content, particularly in terms of compliance and content verification.