Since the advent of Artificial Intelligence (AI), innovators have been accelerating their efforts in its development. Nevertheless, Meta’s CEO, Mark Zuckerberg, is considering a slowdown or even a complete halt in the development of Artificial General Intelligence (AGI) systems that are classified as “high risk” or “critical risk,” according to a newly released policy guideline.

AGI refers to an AI system with the ability to perform any task a human being can do, and Zuckerberg had previously pledged to make such technology available to everyone. However, in a document titled “Frontier AI Framework,” he acknowledges that some highly competent AI systems may not be publicly released due to potential risks.

The framework is primarily concerned with the most significant threats in cybersecurity and risks associated with chemical and biological weapons.

“Our strategy is to prioritize these areas so that we can ensure national security while fostering innovation. The framework provides an overview of the various procedures we employ to foresee and reduce risk when developing frontier AI systems,” according to a press statement about the document.

The framework aims to identify “possible catastrophic outcomes linked to cyber, chemical, and biological risks that we strive to prevent.” It also conducts “threat modeling exercises to predict how various actors might misuse frontier AI to produce those catastrophic outcomes,” and has “procedures in place to maintain risks within acceptable boundaries.” If the company determines that the risks are too high, it will not release the system to the public.

“While the emphasis of this Framework is on our efforts to predict and reduce the risk of catastrophic outcomes, it’s vital to stress that the primary reason for developing advanced AI systems is the enormous potential societal benefits these technologies could offer,” the document states.

However, it appears Zuckerberg is applying the brakes on AGI’s rapid progress towards the future, at least for the time being.