Meta Adopts AI to Transform Moderation on Its Platforms

Meta, the tech giant behind Facebook and Instagram, is accelerating its transition towards automated content moderation. The company aims to significantly reduce the role of human moderators in favor of AI-based systems. This initiative seeks to detect and remove problematic content more quickly before it spreads across its platforms. This change could disrupt the current organization of moderation.

Towards Faster Moderation with AI

Meta plans to gradually decrease the number of human moderators in favor of automated systems, although no specific timeline has been provided. Currently, the company relies on thousands of contractors worldwide to review content reported by users or detected by its internal tools. However, this model could evolve significantly, with a potential reduction of 20% in the workforce, amounting to approximately 15,800 jobs at risk, although Meta has not yet officially detailed the extent of the layoffs.

The company justifies this choice by the promising results of its systems based on language models, which offer increased speed and extensive language coverage. Currently, Meta's tools cover about 80 languages, but the new systems are expected to be capable of processing enough languages to encompass 98% of internet users.

The Ongoing Role of Human Moderators

Despite this technological shift, Meta emphasizes the importance of maintaining human intervention in certain cases. Moderators will continue to handle the most complex and sensitive decisions, such as appeals related to account deactivation or reports involving authorities. Their role will also evolve towards supervision, training, and evaluation of AI systems. Thus, humans are not disappearing but changing their position in the decision-making chain.

This transition, however, raises questions among users, who believe that automated systems can make mistakes, particularly regarding excessive moderation or handling appeals. Meta claims that its new tools will reduce over-moderation errors while improving the detection of the most problematic content.

An AI Assistant for Users

In the short term, Meta is also rolling out an AI-based assistant on Facebook and Instagram. This chatbot is designed to help users manage certain tasks, such as reporting content, recovering accounts, or managing settings. Initially, this feature is being tested in the United States and Canada.

With this evolution, Meta continues its strategy of large-scale automation in a field where technical, economic, and societal stakes remain closely intertwined. Previously, Meta had already thoroughly revised its moderation policies, marking an important step in its overall strategy.