Meta Shifts to AI for 90% of Content Safety and Privacy Reviews

Meta is set to significantly transform its internal safety protocols by replacing a majority of its human-led content and privacy reviews with artificial intelligence. According to internal documents, the tech giant plans to automate up to 90% of the checks that ensure privacy, safety, and risk compliance across its major platforms—Facebook, Instagram, and WhatsApp.

For years, Meta relied on human experts to perform “privacy and integrity reviews,” a process designed to prevent new features from compromising user privacy or promoting harmful content. These manual assessments have played a crucial role in ensuring that products were safe, especially for young users. However, Meta now intends to use AI systems to handle the bulk of these responsibilities, a shift the company believes will accelerate product development and improve efficiency.

Under the new system, developers will complete structured questionnaires about their projects. AI tools will analyze these inputs, identify potential privacy or safety risks, and generate specific requirements for the team to address. Teams will need to confirm that they’ve met these requirements before launching the product.

Although Meta claims human reviewers will still oversee high-risk or complex cases, the majority of low-risk evaluations will now be automated. The company argues this will free up human resources for more critical assessments. CEO Mark Zuckerberg recently emphasized Meta’s growing reliance on AI, noting that within 18 months, the majority of code at the company could be generated by AI.

While Meta promotes this transition as a step toward greater scalability and faster innovation, critics are concerned about whether AI can fully understand the nuanced ethical and safety implications, particularly for vulnerable users. The broader effects on user trust and content integrity remain uncertain as the automation strategy unfolds.