Meta Automates Risk Reviews with AI: Experts Warn of Potential User Safety Concerns
Meta, the parent company of Facebook, Instagram, and WhatsApp, is now using artificial intelligence to automate up to 90% of its internal risk assessments—raising serious concerns among experts and employees.
According to internal documents obtained by NPR, Meta has begun automating key reviews that previously required human oversight. These assessments evaluate the potential risks of new features and updates, including changes to content-sharing policies, recommendation algorithms, and data privacy controls.
The goal, according to Meta, is to speed up product development and improve operational efficiency. But critics worry this shift could weaken oversight in areas critical to user safety, misinformation control, and child protection.
Former Meta employees and insiders told NPR that replacing human reviewers with AI—even partially—could lead to unintended consequences. A former executive warned that faster feature rollouts without proper human checks might increase risks for users globally.
Sensitive areas like AI bias, teen mental health, and misinformation are especially at risk, say internal critics. Despite Meta’s assurances that complex cases will still involve human experts, the heavy reliance on automation has triggered debates about ethics, safety, and responsibility in big tech innovation.
Meta says it remains committed to privacy and transparency while embracing AI to streamline its internal processes. However, experts argue that the balance between innovation and accountability must be handled with care—especially when billions of users are impacted.
