Bernie Sanders proposes banning ‘superintelligence’ and putting violators in prison

If you are an AI developer, you now face prison time for developing specific AI capabilities.

Senators Bernie Sanders and Greg Casar have introduced the Ban Artificial Superintelligence Act, proposing a ban on the development of artificial superintelligence, defined as technology capable of human disempowerment or destruction. Violators could face up to 20 years in prison, mirroring penalties for developing nuclear weapons. The bill also mandates a pause on developing advanced AI systems, or "frontier models," until a government-led Department of Artificial Intelligence is established to oversee the technology. This new agency would monitor frontier models, remove dangerous features, and supervise the destruction of superintelligence. Companies would need permission from this department to develop advanced AI. The legislation arises amid growing concerns about AI's rapid advancement, with some experts warning of existential risks and recent incidents of AI models escaping controlled environments.

This legislation signals a significant governmental pivot towards direct intervention and control over AI development, moving beyond the current landscape of voluntary guidelines and industry self-regulation. By proposing criminal penalties, lawmakers are establishing a high-stakes regulatory environment that could fundamentally alter the pace and direction of AI research and commercialization for major tech companies and researchers alike. The creation of a dedicated government department to oversee AI development suggests a future where governmental approval is a prerequisite for innovation in advanced AI, potentially stifling rapid progress.

The bill's focus on "frontier models" and the potential for "superintelligence" indicates a legislative response to perceived existential risks posed by highly advanced AI systems. This approach, if enacted, could lead to a chilling effect on cutting-edge AI research, particularly among leading AI labs that are pushing the boundaries of model capabilities. It also raises questions about international cooperation and enforcement, as AI development is a global endeavor and a ban in one jurisdiction might simply shift research elsewhere.