Zuckerberg Slams Centralization of AI Power, Backs Open Approach to Counter Anthropic and OpenAI

TechFlow reports, July 29, according to The New York Times, Meta CEO Mark Zuckerberg publicly criticized the closed AI development path represented by Anthropic and OpenAI in an interview, stating that strict control over AI technology will lead to concentration of power, limit public access to this transformative technology, and is akin to "betraying American tech values." Zuckerberg stated that he is skeptical about the path of "building benevolent superintelligence through a single alignment mechanism" and advocates providing "personalized superintelligence" for everyone to adapt to the needs and values of different users.

He also pointed out that technological democratization has always been the general trend in the history of the tech industry. It is worth noting that Meta's own AI strategy has recently shifted—last spring, it invested $14.3 billion to take a stake in ScaleAI and recruited its founder Alexandr Wang, subsequently launching its first paid closed-source model Muse Spark, while continuing to maintain the open-source model line. This statement echoes Nvidia CEO Jensen Huang's recent public strong support for open-source AI, and both parties stand in clear opposition to Anthropic and OpenAI in terms of AI development philosophy.