Why Meta plans to replace 90% of content moderation with AI

Meta is reportedly looking to replace more than 90% of its content review workload with AI by the end of the year for specific content formats, according to a report released by the Financial Times on Thursday. 

The primary motive behind this change is to place the overwhelming majority of the company’s ads and content review under the control of Meta’s Large Language Models (LLMs).

This comes as Meta invests hundreds of billions of dollars into its evolving AI systems. 

The recent revelation of Meta’s AI moderation shift indicated an accelerated implementation that remains highly unpredictable. This major move will provide reputational liability while giving Meta with less control over content and financial judgements. 

AI can make decisions and mistakes at a much faster pace and in higher volumes than humans, amplifying the brand safety risks since autonomous ad buying systems already eliminate human oversight. 

The prominent technology conglomerates reduced alliance on human moderators may compress compliance costs, but it also places more content decisions in the hands of opaque AI systems, eroding control for users and brands. 

It has been observed that the lack of complexity in these automated systems elevates the risk of permanent bans which can immediately isolate brands, challenge established norms, and disrupt customer relationships and media performance. These issues are significant for Mets’s massive social media reach and its importance to brand affinity and engagement strategies. 

Automated content moderation will help Meta to cut costs, but could increase concerns for the marketers who rely on these platforms if the widespread automation doesn't align with improvements in accuracy, transparency, and dispute resolution mechanisms.