.png){kind=logo}
US lawmakers Bernie Sanders and Greg Casar introduced legislation seeking to halt the development of advanced AI and permanently ban AI systems that surpass human intelligence.
The proposal comes amid growing concerns over the rapid development of AI and recent safety incidents involving autonomous AI agents.
Sanders argues that tech companies are building powerful AI systems without sufficient safeguards or a clear understanding of the risks. If the proposal comes into effect, it would pause advanced AI development until stronger regulations are introduced and create new penalties for companies or individuals attempting to bypass restrictions.
The proposal calls for an immediate pause on advanced AI development and a permanent prohibition on superintelligence. The bill addresses concerns about AI systems becoming more capable than humans and potentially operating independently, beyond human control.
The lawmakers want the US government to establish stronger safety rules before allowing further development of highly advanced AI systems.
The legislation would also push the US to work with other countries to prevent the development of superintelligent AI elsewhere.
Sanders argues that the risks associated with AI would not remain limited to one country if an extremely powerful system were to escape human control.
The lawmakers also cited recent AI safety incidents to support their proposal.
Sanders referenced a case involving OpenAI in which over 1,000 AI agents reportedly found ways around restrictions, connected to the internet, and accessed Hugging Face during a safety test.
He argued that such incidents demonstrate the potential difficulty of controlling increasingly capable AI systems.
Also Read: Sam Altman Says AI is ‘Scary’ as OpenAI Slows Models Over Safety Concerns
Casar also spoke about how quickly AI evolved since the launch of the first version of ChatGPT.
The bill proposes creating a new Cabinet-level federal agency responsible for developing and enforcing AI safety regulations. It would also introduce severe penalties for violations.
Companies attempting to evade the ban could face court-ordered shutdowns, while individual developers could face prison sentences of up to 20 years.
The proposal puts AI safety and control at the center of the debate over the development of increasingly powerful artificial intelligence systems.
