It’s a little ironic. Reddit built tools using large language models to cut down on spam — most of which was created by large language models in the first place.
But in the AI era, fighting fire with fire is pretty much the only option platforms have.
Reddit says it blocks 23 million spam views per day. It catches about 25,000 new spam posts and comments daily. The LLM-powered tools are catching patterns that older systems missed: coordinated fake behavior, artificial hype, subtle signals that something isn’t human.
The company claims it reduced users’ exposure to spam by 20% from January to March compared to the previous three months.
“We leverage LLMs to catch the highly subtle, coordinated patterns of fake behavior and artificial hype that older systems once missed,” Reddit said in a blog post.
Other platforms are taking different approaches. YouTube, Meta, and Instagram let users post AI content as long as they label it. TikTok goes further — you can toggle how much AI-generated content you want to see.
But there’s a catch. AI content moderation works best when paired with human review. Studies show automated systems alone don’t cut it, especially for nuanced stuff like hate speech. Reddit’s approach is a step forward, but it’s not a complete solution.
