OpenAI adds text watermarking to ChatGPT and Codex

OpenAI is rolling out an invisible, machine-readable textGrain watermark in ChatGPT and Codex output, initially only for users in the European Union to meet EU AI Act requirements. It won't be a global default at launch, and API customers can opt in to watermarked outputs for select models.

• textGrain watermarking rolls out in the EU only, across all ChatGPT and Codex plans
• OpenAI says textGrain matched or exceeded Google DeepMind's SynthID for text
• API customers worldwide can opt in to watermarked outputs for select models
• Detector access goes to approved researchers, not the public at launch