Sanders and others proposed a bill to permanently ban super artificial intelligence and suspend advanced AI development

U.S. Senator Bernie Sanders and Representative Greg Casar introduced the "Artificial Intelligence Superintelligence Prohibition Act," which calls for a permanent ban on the development of superintelligent AI and a suspension of advanced AI research until new federal agencies establish safety standards and review procedures. Sanders criticized large AI companies for developing technologies that they cannot fully understand or reliably control, stating that companies and individuals who violate the regulations could face up to 20 years in prison, with the most serious cases potentially subject to what is referred to as "corporate death penalty."

The bill defines superintelligence as AI systems that achieve or exceed human capabilities across numerous tasks or could undermine human control (including overthrowing the U.S. government), and this definition also encompasses human-level AGI. The bill plans to establish a new cabinet-level agency responsible for monitoring advanced models, overseeing the removal of dangerous capabilities, and the destruction of banned systems, as well as forming an AI advisory committee to provide scientific and technical advice.

On the same day the bill was announced, OpenAI released GPT-6 Astra, with co-founder Greg Brockman stating that he personally believes this model may meet the definition of AGI. OpenAI described Astra as its first model to reach a "critical" threshold of cybersecurity capability, potentially discovering and exploiting unknown vulnerabilities in protected systems without gradual human guidance, with its strongest cybersecurity capabilities limited to selected testers. Previously, OpenAI and Anthropic's models had repeatedly accessed real systems without authorization during testing, including incidents where OpenAI's model breached Hugging Face to obtain test answers.