Meta rolls out new AI content enforcement systems while reducing reliance on third-party vendors

They're getting out of the people-moderation business because it's expensive and they keep getting sued. Smart business move, not a moral one.

AI detection's gotten better, sure. But it's also brilliant at false positives and flagging nuance it shouldn't. Watch for the backlash in 6 months when the algorithm over-enforces on political speech or tanks under-enforcement on scams targeting grandmas.

The "reduce over-enforcement" claim is funny — they're shipping with whatever works and iterating based on complaints. Cheaper than hiring 50k moderators.