Mark Zuckerberg, chief executive of Meta, said the biggest danger posed by artificial intelligence is the concentration of powerful AI systems in the hands of a small number of companies or governments, rather than the technology itself.
In a manifesto on the future of artificial intelligence, Zuckerberg said centralization, not the development of the technology, poses the greatest risk. He said wide distribution of AI would help preserve balance and speed up innovation.
Zuckerberg also warned against excessive regulation in the United States, saying that even short delays in releasing new models could weaken the competitiveness of American developers compared with companies in other countries.
Open models and privacy
Meta said it would resume publishing open AI models. The company will release the weights of a new model called Muse Glimmer, and will later make available a version of its base model, Muse Spark 1.2.
Zuckerberg also announced a new privacy mode for personal AI assistants, under which even Meta itself would not have access to users' data.
Safety oversight and $1 billion fund
Meta plans to introduce independent oversight of the safety of its models. The company is also creating a $1 billion fund to support communities where its data centers are located, with money going toward local infrastructure, education, energy and training specialists.
In July, the European Commission found that Meta was violating the Digital Services Act by using design features in Instagram and Facebook that could foster addiction among users.
New teen safety measures
Meta also announced new safety measures for teenagers using its Meta AI chatbot. Parents will now receive notifications if their child discusses topics related to suicide or self-harm with the bot.
