On October 7, 2026 Meta announced new artificial‑intelligence tools designed to find ads that act as gateways to child sexual abuse material. The company also reported that it removed 33.2 million pieces of such content from Facebook and Instagram in the first half of the year. Most of the removals – more than 97% – were done before users reported them.
How the new tools work
Meta’s latest large language model – an AI that processes text – looks for “signposting,” meaning ads that appear normal but send users to illegal sites. Instead of just checking the ad copy, the system follows the link to see where it leads and can block the destination if it violates rules. The company also added a “red‑teaming AI agent” that tries to break its own safety systems, helping engineers patch weaknesses before bad actors exploit them. Additional scans are being used to catch exploitation content that older tools missed, and the platform is improving detection of repeat offenders who return with new accounts.
Broader child‑safety push
The new AI measures join a suite of child‑safety features introduced by Meta in 2026, such as parental controls for Meta AI, pre‑teen WhatsApp accounts and Instagram alerts when teens search for self‑harm content. In August, Meta agreed to pay up to $18 billion to settle a lawsuit brought by 29 U.S. states over child‑safety concerns.
Why it matters
For everyday users, the AI could stop harmful ads from silently leading people, especially children, to illegal material. Parents may see fewer risky links appear in feeds, but the effectiveness of the new tools is still being measured, and the company has not released independent audit results.
