Meta hits pause on business terms that would feed linked sites to its AI

Meta just blinked. The company is indefinitely pausing a planned update to its commercial terms that would have forced Meta Business account users to sign off on sweeping AI training using external websites linked from Facebook. A spokesperson confirmed the retreat to Reuters, saying the company is holding off while it keeps talking to stakeholders.

The wording of that statement is doing a lot of work: Meta said it wants to take its time to get this right, and that its existing terms stay in force in the meantime. Translation: nothing changes for business account holders right now, and there's no new date on the calendar.

The update was never subtle. Legal experts flagged the language for Nieman Lab, warning that its vagueness could let Meta brush aside robots.txt preferences — the standard signals sites use to tell crawlers to back off — and hoover up off-Facebook content at scale to feed its AI models.

That's where the money comes in. Practically every news publisher runs a Facebook page, so Nieman Lab raised the alarm that Meta could sidestep millions in licensing fees if the terms went through as written. Publishers and other site owners who thought robots.txt had their back suddenly had a reason to read the fine print.

The pushback was loud enough that Meta tried to calm it down and failed, which led to this pause. The company hasn't killed the idea outright — it says it's still engaging with stakeholders — so the terms could resurface in some form. For now, business account holders keep the terms they already agreed to.

The bigger question hanging over the whole episode is the one critics were already asking: whether Meta expected anyone to skim past language this broad. Meta hasn't answered that one yet.