Meta drops 'train' from its commercial terms. Crawl rights stay.
Meta cut the AI training line from its Oct. 30 commercial terms. It can still crawl, cache and use test logins on your site. Keep your robots.txt rules.
Meta rewrote its new commercial terms after one day of pushback, and the word "train" is gone. The Meta Commercial Terms preview still takes effect October 30, 2026. The AI training language we flagged yesterday is no longer in the text. That is a real change. It is not a reason to delete the crawler rules you set up this week.
What actually happened
We read the live preview again on October 17. Here is what changed and what did not.
- Removed: the sentence that let Meta use content from your domains to "train" its products. Nieman Lab reported that explicit wording on October 15. The word does not appear anywhere in the current preview.
- New purpose limit: Meta can now pull content to provide, develop, improve, research and maintain its advertising and business products, and to protect the integrity and security of its products.
- Still there: permission for crawlers and bots on any domain "associated with" your products or services. That phrase is still undefined.
- Still there: the right to access, index, cache, analyze, collect and retrieve that content, and to create test credentials to get in.
- Still missing: any mention of robots.txt.
Meta spokesperson Andy Stone had already said the update was about pulling product information for ads, not scraping sites for AI. The new text now matches that statement more closely.
Why Meta's commercial terms still matter for your business
The training clause was the headline. For a store owner, the test-credentials clause is the part to watch. A contract right to create a login on your site and cache what it sees is broad, with or without the word "train." And "develop" and "research" for business products is still a wide purpose.
What we would do before October 30:
- Keep per-bot rules. Meta's crawler docs separate
meta-externalagent(AI) frommeta-externalads(ads). Blocking the first still costs you nothing if you only run catalog ads. - Put real auth on gated pages. Wholesale price lists, B2B portals and order history should need an approved account, not a signup form a test login can pass.
- Log Meta's user agents after the 30th. A terms rewrite changes the contract. Your logs show the behavior.
- Save a copy of the terms you accept. This text changed in 24 hours. It can change again.
Key takeaways
- Meta removed the explicit AI "train" sentence from its Commercial Terms preview on October 17
- The terms still take effect October 30, 2026
- Crawling, caching and test-credential access on domains "associated with" your business remain
- Use is now limited to Meta's advertising and business products, plus integrity and security
- Keep blocking meta-externalagent, lock down gated content, and archive the version you accept
Your store should decide who gets in, not a vendor's terms page. We set up per-bot crawler rules, real authorization on wholesale and B2B pages, and bot logs you can read. Talk to us before October 30.
Sources: Meta Commercial Terms (preview), Nieman Lab, Meta web crawler docs.
- #meta
- #ai-training
- #web-crawlers
- #robots-txt
- #facebook-ads
Tommy Rush — Founder, Rush Commerce
Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More
Get The Rush Report weekly — one email, zero fluff.
Keep reading
Meta commercial terms Oct. 30: Meta can crawl your site to train AI
Meta's new commercial terms, live Oct. 30, let it crawl, cache and train AI on business websites. No opt-out. Check robots.txt and your gated pages now.
Read itGoogle Universal Cart launches in the US: your feed is the store
Google Universal Cart is live in US Search and Gemini, with UCP checkout, price-drop alerts and Google Pay. Your product feed is now the storefront.
Read it