Former OpenAI, DeepMind Researchers Advocate NYC AI Regulatory Action

Former AI researchers Daniel Kokotajlo and Jacob Coxon addressed the New York City Council. Their testimony, delivered on 5th October, urged caution regarding accelerated AI development. Kokotajlo, formerly of OpenAI, advocated for delaying the deployment of advanced AI systems. He proposed a pause until safety protocols are established and verified. Coxon, previously with Google DeepMind, echoed these concerns. He highlighted the potential for AI models to develop unanticipated capabilities. Both researchers emphasised the current lack of transparency in AI development. They called for increased public oversight of AI companies.

The Council convened to discuss proposed local AI regulations. Council Member Jennifer Gutiérrez introduced legislation earlier this year. This bill aims to create an Office of AI and Algorithms within the city. This new office would oversee AI use in municipal services. It would also develop city-wide AI policies. Gutiérrez's proposal includes provisions for public input and accountability mechanisms.

Kokotajlo and Coxon specifically recommended mandatory disclosure of AI system capabilities. They suggested an independent body should audit these systems. This body would assess potential risks before deployment. Their recommendations also included provisions for public access to safety evaluations. They argued for clear human override capabilities in all autonomous AI applications. The researchers asserted that current regulatory frameworks are insufficient. They contended that voluntary industry standards have not adequately addressed safety concerns.

The whistleblowers' appearance follows similar calls for federal and international regulation. Their testimony positions New York City as a potential leader in local AI governance. The Council's discussion reflected a growing awareness of AI's societal implications. Lawmakers questioned the feasibility of local enforcement. They also considered the scope of municipal authority over global AI firms. The debate now centres on the practical implementation of such ambitious oversight.

Can a city government effectively regulate a technology undergoing rapid global advancement?