Mark Zuckerberg published a 6,500‑word essay on Monday in which he sketched a future where every person has a personal, all‑knowing AI assistant. He argued that the greatest risk to society is the concentration of advanced AI capabilities in the hands of a few corporations, governments or institutions.
Open‑source AI as a safeguard
In the essay, the Meta Platforms chief championed open‑source AI, saying developers should be able to examine, modify and build on core components. He said the approach would give billions of people "invention superpowers" and enable new businesses, PhD‑level tutoring and personalised lifestyle advice.
To demonstrate the strategy, Meta announced the release of Muse Glimmer, an open‑source model that can run on a personal computer, and promised a more powerful model, Muse Spark 1.2, for developers.
Why concentration matters
Zuckerberg warned that if control of "superintelligence" remains limited to a handful of entities, outcomes will be less favourable for the wider public. He contrasted this with the majority of AI labs, which he said focus on building tools for corporations and governments, thereby shifting power away from individuals.
"Most other labs are focused on building AI for companies, governments, or other institutions, so if those labs lead, then the balance of power will favour larger institutions over individuals," he wrote.
He also urged the United States and its allies to accelerate energy capacity and infrastructure development to stay ahead of rivals such as China, and called for a reassessment of policies that restrict the practice of "distillation", training a smaller model on the outputs of a larger one.
Critics question the pace and motives
Experts warned that the rapid rollout of powerful models carries significant risks. Anthony Aguirre, president and CEO of the AI safety nonprofit Future of Life Institute, asked what makes Meta think it can control technology that is "exponentially smarter, faster, and more capable".
"Meta's chosen path is ultimately one of human replacement and disempowerment, but there is an alternate, pro‑human path, one that brings about individual flourishing and scientific breakthroughs with controllable AI tools," Aguirre said.
Matt Lane, senior policy counsel at the digital‑rights group Fight for the Future, described Zuckerberg as "annoyingly out of touch" but agreed that open‑source AI could improve security against threats such as AI‑driven hacking of voting systems.
What comes next?
Meta plans to make the newly announced models widely available to developers, positioning the company as a leader in open‑source AI. At the same time, regulators in Europe and the United States are expected to scrutinise the implications of broader AI distribution, particularly regarding safety, competition and data protection. The debate over whether open‑source AI will democratise innovation or simply extend Meta's influence is likely to intensify in the coming months.