OpenAI Prepares Invisible Watermark for ChatGPT and Codex Text

OpenAI is preparing to add an invisible watermark to text generated by its artificial intelligence tools, including ChatGPT and Codex.

The planned feature would introduce a hidden marker into AI-generated text rather than a visible label that readers could immediately recognise.

The move centres on identifying content produced by AI tools, as distinguishing machine-generated writing from human work remains a challenge.

Details of how the watermark would be embedded or detected were not provided. It is also unclear whether users would have access to a tool for checking marked text.

No launch date or rollout schedule was specified, leaving the timing and scope of the planned feature uncertain.