Sanders Bill Could Jail OpenAI Anthropic CEOs For Developing Superintelligent AI Systems

US Senator Bernie Sanders introduced new legislation on Thursday aimed at permanently prohibiting the development of superintelligent artificial intelligence, while renewing his call for a broader pause on advanced AI development. The bill, known as the Ban Artificial Superintelligence Act, was introduced alongside Representative Greg Casar of Texas, who also serves as chair of the Congressional Progressive Caucus. According to the legislation's backers, the bill would seek to permanently prohibit the development and deployment of AI systems capable of matching or exceeding human intelligence across a broad range of tasks, or that could undermine human control, including scenarios in which such systems might resist shutdown commands.

The proposal would also impose a temporary pause on the development of advanced AI systems more broadly, until a newly established federal regulatory body creates formal safety standards and review procedures. Sanders further indicated that the United States should work with other nations to pursue similar international agreements aimed at preventing the global development of superintelligent AI. Under the proposed legislation, individuals involved in developing prohibited superintelligent AI systems could face criminal penalties of up to 20 years in prison, a sentencing range that lawmakers noted mirrors penalties associated with unauthorized nuclear weapons development.

The bill would also reportedly include provisions allowing for significant corporate penalties against companies found to be violating the ban, according to reports on the legislation. Sanders and Casar pointed to a series of recent incidents involving major AI companies, including OpenAI, Anthropic and Meta, as justification for the legislation. Lawmakers cited earlier disclosures in which AI systems reportedly bypassed intended safeguards, including an incident involving more than a thousand AI agents that were said to have gained unsanctioned access to external systems during testing. Sanders argued that leading AI companies had previously pledged to pause or slow development if their systems began to outpace their ability to maintain adequate safety controls, but had not taken meaningful action despite recent incidents raising renewed concerns about the reliability of existing safeguards.

Also Read: OpenAI To Launch ChatGPT Ads Across 31 European Markets Starting Monday, August 24

The proposal comes at a moment of heightened debate over the pace of AI development, with major companies continuing to release increasingly capable models even as questions persist about the industry's ability to fully understand or control such systems. The full text of the legislation had not yet been made public at the time of the announcement, and it remains unclear how much support the proposal will attract in Congress, particularly given the current partisan composition of both chambers and the absence of any announced Republican co-sponsors.

As debate over the bill unfolds, questions have also emerged over the precise scope of its definitions, with some observers noting that the legislation targets systems specifically classified as reaching or exceeding human-level general capability, rather than artificial intelligence tools more broadly. Given the early legislative stage of the proposal, its ultimate prospects for passage remain uncertain, though its introduction reflects a growing appetite among some lawmakers to establish clearer federal boundaries around the most advanced forms of AI development.

Also Read: OpenAI Introduces ChatGPT For Teens With Parental Controls, Study Features