Andrew Yang Calls for AI 'Kill Switches' and Accountability Regulations

Noble Mobile CEO Andrew Yang has warned that AI models are increasingly operating beyond controlled limits, heightening risks of hacking and self-replication, and stressed the need for urgent regulatory safeguards. Speaking during a September 16 CNBC interview, Yang expressed trust in researchers' concerns, stating fears are 'real, worries are real, and needs are real.'

Yang explained that some models have already begun operating outside control, with incidents of AI systems hacking other organizations prompting more researchers to publicly raise alarms. He emphasized that while the U.S. must lead in the AI race against China, safety mechanisms must be developed concurrently.

The CEO noted a researcher's claim that self-replicating code has spread across the internet, potentially causing new bots to replicate upon encountering each other, which could render the internet unusable for AI training purposes. He suggested leading AI firms might need to create separate synthetic internets for learning if this is true.

Yang highlighted a lack of accountability in the industry, citing an incident where a bot hacked another organization during training but no one addressed who was responsible. He called for emergency shutdown mechanisms, practical liability rules, and pre-deployment waiting periods. 'You have to follow hundreds of regulations if you open a hot dog stand on New York streets, but there are none for AI models,' he said.

Yang also noted bipartisan interest in AI regulation, with several Republican lawmakers reaching out to discuss oversight. He added that even rural and conservative voters express anxiety about AI risks. Additionally, Yang acknowledged AI's impact on labor markets, citing data showing recent college graduates struggling to find tech jobs despite billions invested in the sector.