Starting October 5, OpenAI lets API customers worldwide opt into textGrain for select models and plans invisible watermarking for eligible ChatGPT and Codex text in the EU over the coming weeks. At a 1% target false-positive rate, detection reached about 80% for 200-token psychology passages and 95% at 400 tokens; replacing 10% of words in 400-token passages cut detection from about 92% to 66%, and replacing 25% cut it to 17%. A watermark cannot identify the user, settle ownership or verify truth, and no detected mark does not prove human authorship.