OpenAI is adding text watermarking in ChatGPT and Codex

OpenAI has announced the rollout of an invisible, machine-readable watermarking system for text generated by ChatGPT and Codex. This feature is currently being introduced specifically for users within the European Union, aligning with the region's evolving regulatory landscape regarding AI transparency. OpenAI claims that its proprietary 'textGrain' watermarking technology performs as well as or better than existing industry standards, such as Google DeepMind’s SynthID, which is also utilized by competitors like Anthropic. By embedding these subtle markers, the company aims to improve the detectability of AI-generated content, helping to distinguish machine-written text from human-authored material. While the initial deployment is limited to the EU, this move represents a significant step in the industry's broader effort to implement technical safeguards for AI accountability and provenance, addressing concerns regarding misinformation and the ethical use of generative models in digital environments.