Yann LeCun: AI sovereignty is a matter of national security

On the one hand, the scientist espouses the need for individual governments not to depend on foreign infrastructures. According to LeCun, the urgency is justified by the fact that the future use of information will take place almost entirely through virtual assistants, making the plurality of models a fundamental democratic pillar on a par with freedom of the press. “I’ve had exchanges with several governments around the world. They all want to ensure their sovereignty when it comes to AI, and I think they’re right,” he said from the stage.

To achieve this goal, however, the solution lies not in closing oneself within one’s own borders, but in transparency. True sovereignty is defended by creating collective and free platforms, such as the Tapestry initiative he promoted: “The only way to get there would be an open and free basic model, on which anyone can create their own specialized assistant”, adapted to their own culture and value system.

On the other side of the fence, LeCun strongly condemns centralized management of technology, rejecting the idea that security is achieved by hiding tools from the public. His criticism is aimed both at the block imposed by the White House on Claude Fable 5 and Mythos 5, and at Anthropic’s internal choice to reserve the use of its cybersecurity systems only to a small circle of large companies.

It is in this elitism that the scientist identifies an authoritarian drift. “There is great arrogance and a superiority complex in the idea that only a few are capable of controlling AI and that the ignorant masses should not have access to it,” he denounced. For the pioneer of the sector, although the risks must be mitigated, limiting access to technology does not represent a real security strategy, but only translates into an exercise of pure political and commercial control.