Meta on a shopping spree: more AI experts change employers

As we all know, Mark Zuckerberg is serious about AI: the head of Meta has poached more top AI experts. They come from OpenAI, Anthropic, Apple and Safe Superintelligence and are to build up the newly founded Meta team Superintelligence Labs. It is not known what the experts were offered in each case, but it is likely to be several million.

With Meta, Zuckerberg has the advantage of being able to cross-subsidize the entire AI business. The money comes in thanks to other business models, especially advertising on social networks. The situation is different for OpenAI, Anthropic or the newly founded Safe Superintelligence, for example, which have so far been dependent on investors. This also means that their options are probably not only more limited, but the entire financial future of the start-ups is rather uncertain.

Daniel Gross is said to be one of the candidates for a change. Together with Ilya Sutskever, co-founder of OpenAI, the early 30-year-old founded Safe Superintelligence, an AI start-up that is dedicated to the safety of AI and wants to develop a superior AI for this purpose. Sutskever is now taking on the role of CEO himself. According to reports, Zuckerberg had actually tried to take over the entire start-up, but is said to have failed because of Sutskever. The situation is different at Scale AI; CEO Alexandre Wang is going directly to Meta, which is acquiring 49% of the start-up, which deals with data processing, for 14.3 billion US dollars.

AI experts from all countries join forces at Meta

Bloomberg reports that Ruoming Pang is also moving to Meta. He formerly worked in the Foundation Model team at Apple and before that at Alphabet. Nat Friedman, former CEO of Github, is also moving to Meta. The programmer confirms at X that he will be developing "amazing AI products" for Meta in the future.

Researcher Yuanzhi Li is joining from OpenAI, while Anton Bakhtin is moving from Anthropic to Meta. A few days ago, it was announced that further OpenAI employees would be joining Meta, including three scientists who had recently set up an OpenAI office in Switzerland and who previously worked at Alphabet.

Recruitment is to be handled directly by Mark Zuckerberg himself, including invitations to his home.

Meta has science and product teams

Meta is taking a multi-pronged approach to AI. On the one hand, there is the newly founded Superintelligence team in Silicon Valley. In addition, the Gen-AI team is responsible for products such as Meta's AI models Llama and Meta AI. The FAIR research team, which is involved in basic research, is also based in Paris, as is Turing Award winner Yann LeCun, who has a kind of special status as a scientist. Even though Llama was originally developed in Paris, other employees are responsible for the downstream product. FAIR is currently working on V-Jepa, for example, a world model that is intended to lead to Advanced Machine Intelligence (AMI).

Unlike OpenAI and Anthropic, Meta primarily offers its models as open models. This is also due to the company's financial capabilities, but also the conviction that open source can be used to develop an ecosystem from which Meta itself can then benefit.

(emw)