# OpenAI will honor a shared robots.txt signal that lets sites opt out of AI training

Canonical: https://theagenticweekly.com/news/openai-robots-txt-ai-training-preference-signal · Evidence: Reported · First seen: 2026-10-30 19:49 UTC · Source published: 2026-10-30 09:30 UTC · Brief published: 2026-11-01 00:21 UTC · Updated: 2026-11-01 00:21 UTC · Author: Agentic Weekly

> OpenAI said on October 30, 2026 that it will honor a shared robots.txt signal for AI training preferences. The mechanism is an IETF Internet-Draft published the same day: site owners write robots.txt groups keyed by a policy token, usage_ai-training, to allow or refuse training per path.

## What changed

- OpenAI's news feed says it will honor a shared robots.txt signal so website operators can tell participating AI providers their training preferences.
- The signal is defined in draft-kelley-aipref-training-preference-signal-00, an individual IETF Internet-Draft published October 30, 2026 and expiring May 3, 2027.
- Sites write ordinary robots.txt groups whose User-Agent line is a policy token, usage_ai-training, then Disallow or Allow paths; the draft's example blocks training everywhere except a press-kit folder.
- The draft says these training preferences are separate from crawl rules, so a site can allow crawling while refusing training.
- The draft is not an IETF standard; its token follows the definition the IETF aipref Working Group has reached.

## In the source's words

> "binds usage preferences (e.g., "no AI training") to paths by writing ordinary robots.txt groups keyed by a per-purpose policy token in the User-Agent line."
> (IETF Internet-Draft draft-kelley-aipref-training-preference-signal-00, https://datatracker.ietf.org/doc/html/draft-kelley-aipref-training-preference-signal-00)

## What we have not verified

- OpenAI's announcement page on openai.com blocks our fetcher; OpenAI's commitment is taken from its own news feed entry.
- We did not check which other providers have committed to the signal, or when OpenAI's crawlers start reading it.

## Timeline

- Source published: 2026-10-30 09:30 UTC
- First seen by our watcher: 2026-10-30 19:49 UTC (via OpenAI News feed)
- Brief published: 2026-11-01 00:21 UTC

## Sources

- [OpenAI News RSS feed (October 30, 2026 entry)](https://openai.com/news/rss.xml) (primary source) (fetched 2026-10-31)
- [IETF Internet-Draft: AI training preference signal](https://datatracker.ietf.org/doc/html/draft-kelley-aipref-training-preference-signal-00) (primary source) (fetched 2026-10-31)

## Related

- [ChatGPT: verdict and setup guide](https://theagenticweekly.com/tools/chatgpt)
- [All news briefs](https://theagenticweekly.com/news)
- [This week's releases](https://theagenticweekly.com/this-week)
- [New AI models this week](https://theagenticweekly.com/models/this-week)

## How to cite

Agentic Weekly, "OpenAI will honor a shared robots.txt signal that lets sites opt out of AI training", published 2026-11-01 00:21 UTC, https://theagenticweekly.com/news/openai-robots-txt-ai-training-preference-signal