Meta is reportedly accelerating plans to replace human content reviewers with AI. The company aims to use AI models to review the vast majority of content and ads by year-end.
Meta said AI now handles around 50% of these review tasks. It wants to push that figure to 90% by the end of 2026. That means nearly all content and ad decisions will soon run through AI. The change spans Facebook, Instagram, and Meta’s other platforms.
This follows a recent incident where around 20,000 Instagram accounts were compromised by hackers. They managed to deceive Meta’s AI support bot, which inadvertently granted them access to accounts that weren’t theirs. However, calling them “hackers” might not be entirely accurate, as this exploit didn’t need any coding expertise or deep system knowledge. The account thieves merely requested that Meta’s AI support system send account verification codes to an email address they controlled.
The core problem lies in how these tools work. People can phrase requests in almost endless ways. That makes it very hard to block every form of misuse. Even when specific queries are blocked, users often find another wording that works.
So the more sensitive tasks Meta hands to AI agents, the bigger the exposure. Each new capability widens the surface for potential abuse. Yet Meta still needs to empower these tools to prove their value.
The company is investing hundreds of billions of dollars into AI. It hopes to eventually sell these systems to other organizations. Demonstrating staff cuts within its own business strengthens that pitch. If Meta cannot show the use case internally, the premise weakens. The open question is whether Meta is moving too fast and opening fresh avenues for exploits.
