In short: robots.txt is the file sites use to state their preferences for how their content is used, including for training core generative AI models. A shared signal in it does not work yet. OpenAI promised to respect it, but gave no date.
Until now every AI provider meant a separate bot and a separate rule. Site operators had to identify the bots one by one and keep a list for each provider. If you want to protect your text from five companies, you write five rules and watch for the sixth to show up.
What OpenAI announced
On 30 October 2026 OpenAI announced it will respect a shared signal in robots.txt. The company calls it a voluntary effort: one stated preference that applies to all participating providers at once. The work builds on the AI Preferences group at the Internet Engineering Task Force (IETF), and OpenAI urges other providers to respect it too. This is a plan: rollout is aimed for "the coming months", and details will come in a new announcement. OpenAI's current bot controls still apply and the new signal does not replace them, it adds on top.
The idea is sound: one signal instead of a list of bots. But there are still gaps. It is not known who the other participants in the working group are, or what the signal will look like in the file. A call to other providers is not a guarantee, and "voluntary" is OpenAI's own word, not a legal requirement. The check is the second announcement: when it has syntax and a date, there is something to add to your file. Until then, leave robots.txt as it is.
The visual is generated code-art, no third-party images.