Bad actors are finding it easier than ever to flood the internet with spam thanks to large language models (LLMs) that can generate content effortlessly. To combat this growing issue, Reddit has developed tools using its own AI to spot and block spam more effectively. In 2023 alone, they've managed to block over 23 million spam views daily.

According to the platform, these new LLM-based systems are particularly adept at identifying subtle patterns of fake behavior that previous methods might miss. Between January and March this year, Reddit users saw a 20% reduction in their exposure to spam compared to the previous quarter.

The irony is not lost on social media experts. While platforms like YouTube, Meta, Instagram, and TikTok allow users to post AI-generated content—provided they disclose it—they still need robust systems to detect it accurately. This means that as these tools improve, they could also be better at identifying harmful content such as hate speech.

However, experts caution that AI content moderation must always be paired with human oversight for best results. The challenge now lies in ensuring that both humans and machines can work together to keep the internet a safe space for all users.