Sanders and Casar introduce bill to ban AI superintelligence, establish federal AI department

Sanders and Casar introduced legislation Wednesday that does two things simultaneously: it proposes banning artificial superintelligence outright and creates an entirely new federal Cabinet department to regulate AI development. The bill's criminal penalty structure, up to 20 years imprisonment, makes this more aggressive than existing tech regulation frameworks, yet the reporting on both sides obscures what the bill actually does versus what it claims authority over.

The verified facts are straightforward. The legislation, introduced in the House and Senate, contains three core elements: a prohibition on developing, training, or operating artificial superintelligence; establishment of a Department of Artificial Intelligence housed at Cabinet level with regulatory authority over advanced AI systems; and creation of an advisory board to guide the department's rulemaking. The bill does not specify how superintelligence will be defined in practice, which the coverage from all three sources leaves unaddressed. MSNBC frames the proposal around the criminal penalties and the word "ban." The Washington Examiner leads with the Cabinet department creation. PBS emphasizes both elements equally.

The left framing emphasizes protection. MSNBC's language choice, "dangerous AI superintelligence" and the 20-year penalty focus, positions this as a defensive measure against an uncontrolled technological threat. The reporting cites concerns about "frontier AI development" but does not establish what current frontier AI systems actually do or whether existing regulatory gaps prevent enforcement. The framing treats superintelligence as a self-evident category of harm, not a disputed technical question. What the left coverage does not foreground is that no superintelligent AI system currently exists, making the ban primarily aspirational rather than preventive.

The right framing emphasizes regulatory scope and bureaucratic expansion. The Washington Examiner's language choice, "Cabinet-level executive agency" and "broad regulatory power," positions this as government overreach into a sector that may not require heavy-handed federal intervention. The reporting emphasizes what the bill creates, not what problem it solves. What the right coverage does not fully explore is whether market-based or sectoral self-regulation has actually failed to address genuine risks, or whether the bill's proponents have documentation of specific regulatory gaps.

What neither side adequately addresses is definitional. Superintelligence remains a contested term in AI research itself. The bill appears to grant the new department authority to define what constitutes superintelligence for enforcement purposes, which means the agency would essentially regulate based on its own categorization. Neither reporting outlet examines the prior attempts at AI regulation, the Biden executive order on AI from October 2023, or how this bill's authority would interact with existing FDA, FTC, and sectoral regulators already overseeing AI applications. The coverage also does not establish whether Sanders and Casar consulted with AI researchers about whether a 20-year criminal penalty is proportional to the alleged harm or whether it would be practically enforceable.

What this actually signals is that Congress is moving toward direct AI governance rather than sectoral regulation, creating a new centralized authority specifically for AI oversight. The political significance lies not in the ban itself, which covers a non-existent technology, but in the institutional precedent of establishing a Cabinet-level department for a single technology sector. The unresolved question is whether that structure will survive committee, and if it does, whether it generates the opposition that broader regulatory frameworks typically encounter.