Meta Delays Its New Terms, but Publishers Still Fear an AI Scraping Loophole
Meta Delays Its New Terms, but Publishers Still Fear an AI Scraping Loophole
Meta’s attempt to rewrite the rules for business accounts has run into a familiar obstacle: publishers who fear that vague platform language can become a licence for AI harvesting.
The dispute began with terms scheduled to take effect on October 30. They said Meta could use content from websites tied to business accounts to “develop, research” and “train” its products — wording publishers read as permission to scrape their sites for AI development. The News/Media Alliance wrote to chief executive Mark Zuckerberg, warning of “severe harm,” while European and British publishers and advertisers pressed regulators to intervene.
After the backlash, Meta removed the explicit reference to training. But critics say that was a cosmetic retreat, not a meaningful safeguard. Cornell law professor James Grimmelmann said the revised terms still leave intact language that could let Meta “scrape entire sites.” Media and intellectual-property lawyer Matt Topic similarly argued that permission to develop undefined “business product[s]” could still encompass AI training.
Meta’s position is that the alarm misreads the policy. Spokesperson Daniel Roberts said the changes are “not about scraping news sites” and do not grant rights that override website owners’ preferences. He pointed to robots.txt, the longstanding tool sites use to tell crawlers what they may access.
On the publishers’ timetable, however, Meta blinked. The company delayed enforcement until November 20, saying it needed more time to explain what is — and is not — changing, including how its web crawlers operate and how robots.txt controls apply.
That postponement eases the immediate deadline, but not the underlying argument. Meta says the safeguards already exist; its critics say contractual language broad enough to permit scraping remains the real problem.
Write a comment