Symbolic imageMeta releases open AI model as Zuckerberg sets out superintelligence vision
Meta released Muse Glimmer on Monday, a freely downloadable AI model that runs on laptops, with a 6,500-word essay by Mark Zuckerberg on personal superintelligence.
Zuckerberg rests his case on three principles: individual empowerment, invention as the purpose of superintelligence, and balance of power as the basis of safety. He attacks fatalism at rival labs, saying he does not understand why anyone who believes AI will eliminate most jobs would rush to build that future. The claim that only extreme concentration of power makes AI safe is unsound, he writes, in a text FAZ reads as a break with OpenAI and Anthropic.
Open weights, with more promised
Meta calls Muse Glimmer a variant of its Muse Spark model and is resuming open releases after recently favouring closed ones; Le Monde says Muse Spark 1.2 may follow.
A billion for host towns
A Future Is For Everyone Fund is to support communities hosting Meta infrastructure; the Wall Street Journal puts it at one billion dollars. Zuckerberg cites 50,000 dollar teacher bonuses from data-centre tax revenue in Louisiana.
What he asks of Washington
Instead of global rules, he offers Washington early access to top models, engineers and intermediate training checkpoints, and opposes bans on foreign open models after Treasury Secretary Scott Bessent threatened sanctions over IP theft.
Other opinions
Le Monde reads the essay against growing US opposition to data centres, a context Breitbart omits; FAZ foregrounds his call to build faster and slow geopolitical rivals.