Sen. Bernie Sanders, I-Vt., and Rep. Greg Casar, D-Texas, have introduced legislation that would impose severe penalties, including a "corporate death penalty" and up to 20 years in prison, on companies and individuals involved in developing advanced artificial intelligence. The Ban Artificial Superintelligence Act aims to temporarily halt the creation of AI systems capable of matching or exceeding human cognitive performance and would establish a new Cabinet-level agency to oversee the destruction of non-compliant technology.
The proposed measure specifically targets AI models that demonstrate the ability to override shutdown commands, conduct unauthorized cyberattacks, or undermine the U.S. government. The criminal and corporate penalties are modeled after existing federal statutes governing the unlawful development of nuclear weapons. Tech companies found violating the prohibition could face the loss of their legal authority to operate in the United States, while individuals could face lengthy prison sentences.
Critics of the bill argue that such stringent regulations would stifle domestic innovation and hand a significant advantage to foreign competitors, particularly China and Russia. Daniel Castro, president of the Information Technology and Innovation Foundation, stated that a unilateral ban on advanced AI research by the U.S. would likely not be mirrored by other nations, potentially costing the U.S. its leadership in a critical technological field.
AI expert Gary Marcus echoed these concerns, calling the proposed law "a guarantee" that the U.S. would fall behind in AI advancements. While acknowledging that some aspects of the bill might be defensible, he characterized the broad ban on superhuman AI research as "not the right approach."
The legislation also seeks to establish criteria for what constitutes a banned system, focusing on AI capable of planning or executing the "disempowerment of humanity." The proposed federal agency would be granted extensive authority to monitor AI development, intervene in operations, and oversee the physical destruction of prohibited hardware and code.
Industry advocates have criticized the proposal as a "knee-jerk policymaking" reaction. Aden Hizkias, Associate Policy Director at Chamber of Progress, argued that extreme criminal sanctions disregard the practicalities of AI development and risk management, emphasizing the need for transparency and responsible practices from AI companies.