Meta moderation shifts to AI: 98% language support

Meta, content moderation radically changes its approach. Gradually the majority of human moderators AI systems is preparing to replace it with . The company states that with this transition, it will be able to intervene faster and detect the most serious violations more accurately. This change will have an impact that users will directly feel, especially in account objections and content removal processes.

Meta states that the early results of the large language model-based moderation tools that it has been testing for a long time are positive. These systems go beyond existing infrastructure and cover a much wider range of languages. Current moderation tools support approximately 80 languages. However, new AI models can understand the languages ​​used by 98 percent of internet users.

The company emphasizes that human moderators will not be completely disabled. Expert teams; It will continue its mission to train AI systems, evaluate performance and make the most critical decisions. Human oversight will continue in areas such as account closure objections and sensitive cases forwarded to law enforcement.

In the short term, a concrete change is starting on the user side. Meta is a platform that works within Facebook and Instagram. AI support assistant started to present. Therefore, this system can perform operations such as creating a content complaint, managing account objections, resetting your password and editing account settings directly on your behalf. The company initially offers this support on a limited basis in the USA and Canada for users who have lost account access.

Meta AI expands moderation system globally

Meta will reshape the job descriptions of thousands of contracted moderators during this transformation process. However, it does not share the extent of the decrease in the total workforce. The company states that AI-based systems make fewer erroneous interventions and capture the most serious violations with higher accuracy. In contrast, concerns remain among users about the error rate of automated systems and the difficulty of objection processes.

Meta notes that the rules themselves do not change, only transforming how those rules are applied. This approach can directly impact the way content moderation on the platform is perceived.