Traditionally, spam campaigns were designed to influence human readers through fake reviews, misleading comments or promotional posts. Today, an increasing amount of spam is being created not just to persuade people directly, but to influence the information that AI systems retrieve and cite. In other words, the target is no longer only the user scrolling through a forum but also the AI chatbot that may later repeat that content as part of its answer.
The growing challenge is highlighted by Reddit in a recent trust and safety update, where the company stated that its AI-powered moderation systems now prevent around “23 million spam views every day”, detect approximately “25,000 spam posts and comments daily”, and revoke nearly “2 million inauthentic votes”. According to the company, these efforts helped reduce spam exposure for users by “20 per cent year-on-year during the first quarter”. Reddit described these measures as part of routine platform maintenance, but they also illustrate how protecting online conversations has become increasingly important in the generative AI era.
AI moderation becomes the next battleground
According to Bloomberg, marketers are increasingly experimenting with “generative engine optimisation (GEO)”, the practice of creating online content to influence responses generated by AI chatbots. Rather than trying to rank highly on traditional search engines alone, brands now seek mentions that AI systems may later surface as recommendations.
This changes the objective of spam campaigns. Instead of maximising clicks from human users, some campaigns now attempt to insert promotional material into conversations that AI systems may later treat as genuine community discussions.
Reddit said its newer moderation systems increasingly identify these coordinated attempts before they gain visibility. The company noted that it analyses signals from the moment an account is created to stop suspicious actors before they begin posting.
The platform also said it now uses large language models to detect “highly subtle, coordinated patterns of fake behaviour and artificial hype” that previous moderation systems often failed to identify.
A widespread challenge
The Wikimedia Foundation has said that human-created knowledge is becoming more valuable as AI companies increasingly rely on public information to train and power their models. In its 2025–26 Annual Plan, the Foundation warned that the internet is seeing more low-quality AI-generated content, making reliable, community-created information a “precious commodity.”
Discord also uses machine learning to proactively identify harmful behaviour, spam and coordinated abuse before it reaches users. According to the company’s safety documentation, these AI-powered systems work alongside human reviewers to strengthen community moderation.
X has expanded AI’s role through AI-generated Community Notes, but requires every AI-written note to be reviewed and rated by human contributors before it is published. The company says the approach is designed to scale moderation while maintaining the credibility of information shared on the platform.
The rise of generative engine optimisation
The changing nature of online spam is closely linked to the emergence of GEO. Borrowing from traditional search engine optimisation, GEO focuses on increasing the likelihood that AI systems will cite particular content while answering user queries.
According to Bloomberg, agencies have already emerged that specialise in helping brands improve their visibility in AI-generated responses. One example is ReachLLM, a startup that creates online content, including Reddit posts, to increase the chances that ChatGPT and other AI assistants will cite those discussions.
According to Bloomberg, citing ReachLLM founder Shanzila Ahmed, some posts have been cited by ChatGPT within a day of publication. At the same time, she acknowledged that Reddit has removed some of the same posts as part of its spam enforcement efforts.
The situation illustrates an ongoing cycle in which marketers develop new optimisation techniques while platforms strengthen moderation systems to detect increasingly sophisticated promotional behaviour.
Although the field remains relatively new, these findings indicate that protecting online discussions has implications beyond individual communities. Manipulated conversations may eventually influence the information presented by AI assistants to millions of users.
Why trust and safety are becoming core infrastructure
As AI-generated content becomes easier and cheaper to produce, trust and safety functions are taking on greater strategic importance for digital platforms. Historically, moderation was often viewed as an operational requirement focused on enforcing community rules and removing harmful content.
If AI assistants rely on public discussions to answer questions, the quality of those discussions becomes increasingly valuable. Platforms, therefore, face pressure to ensure that conversations remain representative of genuine user experiences rather than coordinated promotional campaigns. Reddit’s latest disclosures highlight this shift.
According to Reddit, community moderators were responsible for more than 52 per cent of all post and comment removals between July and December 2025, demonstrating that AI moderation complements rather than replaces human oversight.
An evolving challenge
The relationship between AI-generated spam and AI-powered moderation is likely to continue evolving. As marketers experiment with new techniques to improve visibility inside AI-generated answers, platforms will continue refining detection systems to distinguish authentic user contributions from coordinated promotional campaigns.
Reddit itself acknowledged that this remains an ongoing process rather than a problem that can be fully solved through automation alone. The company said it continues to combine AI models, behavioural analysis, human moderators and community reporting to identify emerging forms of abuse.
For online communities, AI is becoming an additional layer of defence that supports human moderators in identifying coordinated abuse and maintaining trustworthy conversations.
