Zuckerberg: AI's biggest risk is one entity with too much control

Meta CEO Mark Zuckerberg defended artificial intelligence (AI) in a 6,500-word manifesto, arguing that common concerns are overstated and fail to consider the technology's potential benefits. The manifesto suggests the biggest risk is excessive power concentrated in the hands of a single government or entity, rather than specific risks often discussed.

Zuckerberg envisions a highly personalized AI that empowers individuals and stresses the importance of global leadership in AI development. He believes that the values of liberty, open inquiry, free enterprise, and equal opportunity should guide the future of AI, ensuring its benefits are shared by all. Zuckerberg warns that excessive intervention by the U.S. could grant an advantage to foreign regimes like China, particularly if technology lead times are shortened.

To address this, he proposes making AI tools accessible to everyone, whether via free or affordable versions with dynamic auction pricing for compute resources. Meta's future AI agents will offer enhanced privacy protections, ensuring users' information remains private from even Meta or other service providers. The company plans to resume open-source AI model releases, with Meta Superintelligence Labs leading this initiative.

Zuckerberg also announced that Meta will maintain an independent board overseeing model-release safety criteria and review processes. To mitigate community backlash over data center development, Meta is launching the Future is for Everyone Fund, committing $1 billion to U.S. communities where data centers operate. The fund will support teachers, first responders, and energy and water infrastructure.

Additionally, Meta will launch the America's Workforce Academy, a training program offering guaranteed jobs at its data centers for workers in skilled trades.