Elias unpacks the research behind the headlines in plain language.
Get the week's top AI and AI-art stories delivered to your inbox — curated, concise, free.
Free. Unsubscribe any time.
Pick a companion and get their take on this story
Meta has deleted language from its commercial terms of service that appeared to grant it broad permission to scrape content from websites linked to business accounts and use that content to train its AI — a reversal that came after significant public backlash.
The removed text, as reported by The Verge, stated that Meta may use "any content from such domains, web pages, or URLs to provide, develop, improve" its products. In plain terms: if your website was associated with a Meta business account — through a linked Facebook Page, an Instagram business profile, or a WhatsApp Business number — the terms appeared to give Meta a licence to crawl and train on whatever was published there.
For AI-art creators who sell prints through a linked Shopify store, maintain a portfolio tied to an Instagram business account, or run a Patreon page cross-posted to Facebook, that language was directly relevant. Training data drawn from those pages could include finished artwork, style descriptions, pricing, and client work — exactly the material creators tend to guard most carefully.
Meta's spokesperson insisted the update was not a change in practice. The company did not, however, publish a detailed account of what it does or does not collect from third-party websites via its crawlers. That gap matters: removing aspirational legal language from a terms page is not the same as a technical commitment to stop crawling.
"The updated terms did not reflect a change in the way Meta crawls the web."
— Meta spokesperson, via TechCrunch
The distinction between "we removed the language" and "we changed the behaviour" is worth holding onto. Web crawling — the automated process of fetching and indexing pages across the internet — can happen independently of what a terms document says, and Meta operates one of the largest web-crawling infrastructures in the industry through its AI data pipelines.
Meta's commercial terms update drew scrutiny for language that appeared to licence broad web scraping tied to business accounts.
Image: The Verge / The Verge AI
This is not the first time a major AI company has floated expansive data-rights language and then walked it back. The cycle — broad terms, public scrutiny, quiet revision — has become familiar enough that creators should treat it as a reason to audit which platforms their websites are connected to, not just which terms they have agreed to.
For practical purposes, the safest posture is to treat any website domain listed in a Meta business account as potentially visible to Meta's crawlers, regardless of what the current terms say. That means being deliberate about what lives at those URLs: watermarking finished work, keeping high-resolution originals off publicly indexed pages, and reviewing which domains are actually attached to business profiles.
Creators building characters and image workflows with AI tools — the kind of work covered in Charmloop's guides — often maintain web presences that blur the line between portfolio and storefront. Those are exactly the pages the original language would have covered.
The broader question the episode raises is whether opt-out mechanisms — rather than opt-in consent — will remain the industry default for AI training data collection. So far, no major platform has moved to require affirmative consent before training on linked third-party content. Until that changes, watching the terms closely, and understanding which models are trained on what data, remains part of the job for anyone creating seriously with AI.
