A preview of Meta's commercial terms, effective October 30, 2026, grants it the right to crawl websites tied to businesses that use its ad and business tools. After backlash it dropped the word "train" — but critics say training was narrowed, not banned.
2026-10-17 · by Moe Ameen
In October 2026, a preview of updated Meta commercial terms — the agreement that governs businesses using Facebook, Instagram, WhatsApp, and Messenger ad and business tools — surfaced with a new clause granting Meta the right to use automated software to access, index, cache, copy, and re-crawl websites associated with those businesses. Legal experts and press reviewing the preview warned that because most online publishers and brands also run Meta ads or business pages, the language appeared to let Meta crawl their sites, and an earlier version listed using that content to "train" Meta's products. The revised terms are set to take effect October 30, 2026.
After pushback from business customers and news organizations, Meta rewrote the clause in mid-October 2026, removing the word "train" and tying the crawling to its "advertising and business products" plus the integrity and security of its services. Meta told reporters it had "heard" the concerns and wanted to "clarify," said the crawling terms only ever applied to its ads and business products, and said it was not changing how its web crawlers work. The company's position is that content owners can still manage access through robots.txt; its documented crawler identifies as Meta-ExternalAgent.
Critics say the edit is narrower than it looks. The News/Media Alliance's Sam Quigley said training is "still potentially on the table" because it was dropped from the listed uses but not expressly prohibited, and Cornell law professor James Grimmelmann read the pre-revision language as clear permission for Meta to train AI on site content. Reports also note there is no opt-out from the terms themselves: businesses accept them passively by continuing to use Meta's commercial products after October 30, and a company can't reject the terms while staying on the services. Blocking crawlers via robots.txt is a separate technical step — and experts caution that a contract a business accepted could be argued to outrank a text file on its own server. Coverage is very recent and sources disagree on specifics, so Meta's live Commercial Terms and Self-Serve Ad Terms are the pages to check before relying on any one reading.
This is a rights-and-crawling story, not a content tool launch, so be clear about what [Kompozy](/) is and isn't: it does not block crawlers, edit your robots.txt, or opt you out of anyone's terms — disallowing Meta-ExternalAgent on your server is a job for your site config, not a content engine. What Kompozy addresses is the second-order problem underneath the headline. When your owned site is simultaneously being crawled more and sent less traffic, leaning harder on that single property is the fragile move; the hedge is to own distribution across many surfaces at once.
Concretely: take the article, transcript, or product update you'd normally bury on a page a crawler indexes, drop it into Kompozy's Quick Ingest, and let it fan out into formats that live where your audience actually is — a captioned [Persona Shorts](/glossary/persona-shorts) avatar video, a brand-exact [Carousel](/glossary/hyperframes), Quote Graphics, Photo Posts, a Blog Article, and an Email Newsletter, all held to one voice by the [Persona Brief](/glossary/persona-brief). [Autopilot](/glossary/autopilot) then schedules and publishes the set across the eight social platforms plus blog and email behind a per-post review. The point isn't to feed or starve a crawler — it's that your reach stops depending on whether a single site gets indexed and summarized, because the same source is now a week of native posts you published and own. For the policy side of the story, pair this with how publishers are responding to crawling and AI traffic loss in our coverage of [Cloudflare's block-training-allow-search controls](/news/cloudflare-block-ai-training-allow-search) and [Google's publisher tools](/news/google-publisher-controls-ai-traffic-loss).
A preview of Meta's commercial terms, effective October 30, 2026, adds a clause letting Meta use automated software to access, index, cache, copy, and re-crawl websites associated with businesses that use its ad and business tools. Because most publishers and brands run Meta ads or pages, legal experts warned it appears to cover their sites.
It removed the word "train" from the listed uses after backlash in mid-October 2026 and tied the crawling to its "advertising and business products." Critics — including the News/Media Alliance and Cornell law professor James Grimmelmann — say that narrows the stated uses but doesn't expressly ban training, so it remains "potentially on the table." Check Meta's live terms for the current wording.
There is no opt-out from the terms themselves; businesses accept them by continuing to use Meta's commercial products after October 30, 2026. Meta says site owners can still manage access via robots.txt, where its crawler identifies as Meta-ExternalAgent — though experts note a contract a business accepted could be argued to outrank a robots.txt file.
Kompozy doesn't block crawlers or change your robots.txt — that's a site-config task. It addresses the deeper issue: when your owned site is crawled more and sent less traffic, Kompozy turns one source into finished posts published natively across the eight social platforms plus blog and email, so your reach isn't pinned to a single page a crawler indexes and an AI answer summarizes.