00:00In India, Meta has a massive hate speech detection problem. Just look at these numbers. Globally,
00:06when Meta takes action against hate speech on Instagram, around 90% of that content
00:10is detected by Meta itself before a user reports it. In India, in the first half of 2026,
00:17it's less than 30%. And Facebook shows a similar gap. But here's the thing, Meta's systems used
00:23to catch much more hate speech in India on their own. In June 2025, Meta proactively caught
00:2982% of the hate speech it acted on in India. One year later, it was just around 25%.
00:35And India isn't some small market. It's actually Meta's largest user base in the world.
00:40So why did Meta's hate speech detection in India dropped so dramatically?
00:44In early 2025, Meta announced a broader shift in its moderation strategy, focusing its automated
00:50systems on illegal and high severity violations while relying more on user reports for less
00:57severe ones. But that doesn't explain why the drop in India was so dramatic.
01:01Another possible issue is local context. AI-based content moderation can struggle to understand
01:06how content is interpreted across different languages and cultures. So whatever is driving
01:11the drop, the result is pretty striking. In India, Meta is now relying much more heavily on users
01:18to flag hate speech that its own systems don't catch first.
Comments