Overview
- Mark Zuckerberg published a manifesto that promises personal superintelligence agents and says Meta will resume releasing some open-weight or open-source models to make advanced AI more widely available.
- Researchers and civil-society experts argue the plan is unlikely to help the most marginalised because many places lack reliable electricity, affordable devices, high-speed connectivity, and local compute to run large models.
- Critics point to recent pauses on projects such as a Microsoft and G42-backed geothermal data centre in Kenya and warn that data centres create few long-term local jobs while adding environmental and grid pressures.
- Open-source advocates and analysts also question Meta’s credibility, citing limited transparency, a custom license that restricts reuse by other big firms, and models that lag peers on coding and reasoning tasks.
- The debate could reshape policy and markets because it raises hard questions about who sets standards for safety and access, how governments should invest in local AI capacity, and how tech rivalry will affect developing countries.