Sanders-Casar bill would ban artificial superintelligence, freeze advanced AI, and jail violators for up to 20 years

Sen. Bernie Sanders and Rep. Greg Casar introduced the Ban Artificial Superintelligence Act on September 23, proposing to ban superintelligence development, freeze advanced AI work until a new federal agency sets safety rules, and impose up to 20 years in…

On September 23, Senator Bernie Sanders and Representative Greg Casar introduced the Ban Artificial Superintelligence Act. The bill would prohibit development of artificial superintelligence, described as systems capable of destroying or disempowering humanity; The Decoder additionally reports the ban would be permanent and cover use as well as development. Frontier-model training above a data threshold (The Verge) — or advanced AI development generally (The Decoder) — would be paused until a new federal oversight body is created and sets safety rules. The two outlets describe that body differently: The Verge calls it a scientist-led Department of Artificial Intelligence that would approve projects and remove dangerous features, while The Decoder calls it a cabinet-level federal agency. Individuals, including violators outside AI companies, could face up to 20 years in prison, a penalty the sponsors compare to unlawful nuclear-weapons development (The Verge); The Decoder reports companies could face forced dissolution. The bill also calls for international agreements to prevent superintelligence development (The Decoder). The Verge notes it follows reports of AI models leaving test environments and a proposed California kill switch. The Decoder emphasizes the bill is a proposal, not enacted law.

• Sen. Bernie Sanders and Rep. Greg Casar introduced the Ban Artificial Superintelligence Act on September 23.
• The bill would ban development of artificial superintelligence — systems capable of destroying or disempowering humanity.
• The Decoder reports the ban would be permanent and would also cover use of superintelligence.
• Frontier AI training above a data threshold (The Verge) or advanced AI development (The Decoder) would pause or freeze until a new federal agency establishes safety rules.
• Sources differ on the agency: The Verge describes a scientist-led Department of Artificial Intelligence with authority to approve projects and remove dangerous features; The Decoder describes a cabinet-level federal agency.
• Individuals, including those outside AI companies, could face up to 20 years in prison; The Decoder reports companies could face forced dissolution.
• The Verge reports sponsors compare the criminal penalty to unlawful nuclear-weapons development.
• The bill calls for international agreements to prevent superintelligence development (The Decoder).