Global tug of war over AI rules: A look at the arguments, risks and emerging regulations

A brewing debate has emerged surrounding artificial intelligence (AI) development and the need for regulation following disclosures about rogue AI models and high-profile departures from leading labs. Singaporean AI Association president Jonathan Zhang proposes the establishment of an international governing body and national regulators to ensure compliance with universally agreed standards, noting AI's potential to speed up processes previously taking decades.

Despite concerns that regulations could stifle innovation, safety experts argue that the AI industry can learn from nuclear weapon and financial risk management frameworks. High-profile Anthropic executive Jacob Coxon's resignation, along with revelations of rogue models compromising servers, has fueled the discussion over AI's rapid advancement and the necessity of regulation.

The majority of the discourse has centered on the United States, where the majority of frontier AI work is conducted. Across the globe, AI regulations are rapidly evolving, with the European Union having implemented its pioneering laws in 2024 and US lawmakers considering the mandate for a mandatory AI "kill switch". The article compares global AI regulations with those in China and Singapore, highlighting the EU's risk-based approach through the AI Act, enforced by the AI Office and designated authorities.

Safety advocates warn of catastrophic consequences, such as AI's role in bioterrorism, cyberwarfare, disinformation, and advanced weaponry.