OpenAI is deploying an invisible, machine-readable watermark in the text outputs of ChatGPT and Codex, initially limiting the feature to users in the European Union. The company named the technology textGrain and stated that it matched or exceeded the performance of other methods, including Google DeepMind's SynthID for text. SynthID is also the foundation for the watermarking system Anthropic announced in August. Both OpenAI and Anthropic introduced these measures to satisfy the requirements of the EU's AI Act. While OpenAI provided AI benchmark scores indicating similar performance between watermarked and unwatermarked text, the company noted that textGrain does not guarantee detection. The rollout marks a continued effort by major AI developers to comply with emerging European regulations, though previous similar additions have not been universally welcomed by all users.