Facebook targets AI slop and copycat content, rewarding originals

Facebook has announced a major feed cleanup aimed squarely at AI junk, stolen videos and recycled material, warning that accounts reposting others' work or relying on low-effort AI material will see reach and revenue hit. The company says enforcement so far this year has taken down some 10 million fake accounts plus more than half a million spammers, with promises of clearer explanations when a post is down-ranked.

The move came days after YouTube clarified its own policy on unoriginal, mass-produced content, a coordinated signal that the platforms are drawing a line against the flood of AI-generated filler. For genuine creators drowned out by copy-paste operations, the shift is being welcomed as overdue.

Separately, Meta fixed a serious bug in its AI chatbot platform that let technically savvy users view strangers' private AI prompts and their generated replies simply by guessing prompt ID numbers. The researcher who reported it received a bounty. The incident is a reminder that fast-moving AI features can expose sensitive data, and that businesses running chatbot experiences should audit their own security practices.

On the feature side, Facebook now lets users add music to text posts and choose AI-generated background images, and Threads is testing sign-up via Facebook accounts instead of requiring Instagram, tapping the larger user base to fuel growth.

Content benchmarks of the week: video remains the top engagement format across Instagram, TikTok and Facebook, though text-only posts surprisingly lead on LinkedIn, and analysts are tracking eight social trends expected to reshape the next few years, with AI touching nearly all of them.