Invisible signals in AI text
Researched retrospectively on 13 October 2026. OpenAI explains its text watermarks and announces an advertising test. Both affect how we interpret AI output.
AI text gets an invisible provenance signal
On 5 October, OpenAI described a text watermark: a subtle statistical pattern in word choices that may indicate its origin. API customers can opt in for selected models. A gradual rollout is planned for eligible ChatGPT and Codex text in the EU. Short, edited or translated text can be harder to detect. Detection does not establish accuracy or identify the author.
What it means: Our interpretation: A provenance signal cannot replace checking the claims.
Example: A possible use: An editor checks a passage and separately verifies its sources.
- OpenAI · Our approach to EU text provenance rules · 5 October 2026
Also worth knowing
A visual advertising test is announced for ChatGPT
OpenAI announced a test of visual ads during image generation. It is planned for later in October in the US on Free and Go. The company says ads remain labelled and separate from generated images.
What it means: Our interpretation: Paid placements should remain easy to recognise.
Example: Hypothetically, a labelled furniture ad appears beside a living-room idea.
- OpenAI · Building advertising for the way people use AI · 5 October 2026