Lawmakers have introduced sweeping federal legislation designed to restrict the development of artificial superintelligence and establish unprecedented government authority over AI advancement. According to AI Weekly, Senator Bernie Sanders and Representative Greg Casar unveiled the Ban Artificial Superintelligence Act on September 23, marking one of the most aggressive policy approaches to AI governance introduced in Congress.
The proposal targets AI systems that would match or exceed human-level cognitive performance across a broad range of tasks. Under the legislation, creating such systems would face a permanent prohibition, with companies found in violation facing either corporate dissolution or criminal sentences of up to 20 years for responsible executives. The penalties are calibrated to rival those imposed for nuclear violations, reflecting the lawmakers' assessment of the risks posed by uncontrolled AI development.
Regulatory Pause and New Federal Authority
Beyond the superintelligence ban, the legislation would impose a temporary pause on advanced AI development until federal safety standards are established. The proposal creates a cabinet-level Department of Artificial Intelligence tasked with enforcing these standards and monitoring cutting-edge AI research activities.
The creation of a dedicated federal department signals a fundamental shift in how policymakers view AI governance. Rather than distributing oversight responsibilities across existing agencies, the proposal consolidates AI regulation under a single authority with cabinet-level resources and authority. This structure would grant the department significant power to define what constitutes prohibited development and to investigate potential violations.
Context and Industry Implications
The timing of the proposal reflects intensifying debates within Congress over how aggressively to regulate artificial intelligence. As large language models and other AI systems become more capable, lawmakers face pressure from both safety advocates warning of existential risks and technology companies arguing that overregulation could harm innovation and competitiveness.
- The superintelligence ban targets systems exceeding human performance in most cognitive domains
- Criminal penalties of up to 20 years would apply to corporate executives responsible for violations
- A development pause would remain in effect until federal safety frameworks are finalized
- The new AI department would possess cabinet-level status and enforcement authority
The Sanders-Casar legislation represents a particularly restrictive approach compared to other regulatory proposals circulating in Congress. While some lawmakers have advocated for transparency requirements, liability frameworks, or sectoral regulation, this bill takes the position that certain categories of AI development should be prohibited entirely rather than regulated.
Industry response to such proposals has been mixed. Some safety researchers support broad restrictions on superintelligence development, citing concerns about uncontrolled capability advancement. Technology companies and investors, conversely, have warned that overly restrictive regulations could push AI development offshore or create compliance costs that disadvantage smaller competitors.
The likelihood of the proposal advancing through Congress remains uncertain, particularly given divisions within the technology sector and questions about enforceability. However, the introduction of such legislation signals that superintelligence regulation and centralized AI governance are now serious topics in legislative discussions.



