- Reddit says that it blocks 23 million spam messages per day that would otherwise be viewed by users, with the figure reaching almost 25,000 new spam posts and comments per day.
- Platform says that using an LLM for reviewing potentially fake content may have reduced spam by 20% in a single–month period compared with the previous three months.
- The material also notes that it is believed that the moderation of content using the LLM is carried out with the participation of human users, and the marketing analytics in particular are used to verify the effectiveness of the changes.
Regarding the material on TechCrunch, The Verge, The Next Web.
Reddit says that it uses large language models (LLMs) to combat spam that would otherwise be seen by users. The platform says that it blocks 23 million spam messages per day that would otherwise be viewed, reaching almost 25,000 new spam posts and comments per day.
The material also says that the new system helps Reddit users view fake content less, and that it has reduced spam by 20% compared with the previous three months.
In addition, the report notes that the platform may have fewer spam messages, but the moderation of content with the participation of the LLM is carried out with human review. The material also says that the marketing analytics are used to verify the effectiveness of the changes.
According to The Verge, the industry SEO in AI prompts is being practiced with the aim of influencing systems. In particular, as described in the material, AI platforms may generate content with hidden advertising, which in turn can influence search results; it also suggests that Google may be using this for the promotion of its own Gemini.
Meanwhile, The Next Web writes that marketing technologies are used to increase Reddit users’ engagement, with a focus on posting and commenting, which makes it possible to create spam messages that ChatGPT and Gemini can generate. The Next Web also suggests that Reddit is using moderation systems to detect and block such content.