What happened
On 23 September 2026, Sen. Bernie Sanders (I-VT) and Rep. Greg Casar (D-TX), chair of the Congressional Progressive Caucus, formally introduced the Ban Artificial Superintelligence Act. The bill would permanently prohibit AI 'superintelligence' (systems exceeding human cognitive ability or capable of destroying/disempowering humanity), pause development of advanced AI pending creation of a new cabinet-level Department of Artificial Intelligence with explicit rulemaking, and require federal pre-approval for advanced AI deployment, with criminal penalties up to 20 years in some cases. Multiple current employees at OpenAI, Google DeepMind and other labs endorsed it.
Why it matters
This is the most sweeping frontier-AI proposal yet in the 119th Congress, from the chair of the Congressional Progressive Caucus, and it anchors the upcoming midterm and 2028 debate on AI governance. While passage is unlikely in the Republican-controlled Congress, it sets the progressive baseline for AI-safety legislation and signals the direction of state and multilateral discourse on pauses and superintelligence bans.
Action needed
Frontier AI developers and data-centre operators should monitor the bill's progress; if enacted it would create a federal pre-approval regime and mandatory pause, and its policy framing will influence state and industry safety commitments.