Meta CEO Mark Zuckerberg has unveiled the company’s bold new vision for AI, centered around building “personal superintelligence” to empower individual users rather than automating all tasks. According to Meta’s official announcement, the aim is to create AI assistants that work toward personal goals through multimodal, context-aware devices like smart glasses. In contrast to many industry players racing to automate workflows, Meta’s approach emphasizes enhancing user capability. Zuckerberg pointed to “personal devices like glasses” as the future's primary computing interfaces, highlighting their ability to provide rich context and more meaningful AI experiences. A key shift in strategy is Meta’s stance on open-source AI. The company signaled it may be “careful” about releasing its most advanced models due to safety concerns surrounding artificial superintelligence. This is a departure from its earlier open-source-first stance. The decision comes alongside reports that Meta paused development of its “Behemoth” model to redirect focus toward closed systems and smarter, safer AI pathways. Why it matters: As China accelerates with open-source breakthroughs like GLM-4.5 and Wan2.2, Meta’s pivot to a more closed and personalized ecosystem marks a significant divergence. The strategy plays well with Meta’s investments in smart glasses and XR platforms, potentially giving the company a unique edge in delivering personal AI experiences safely. With safety and personalization in focus, Meta is setting itself apart in the global AI arms race.