AI safety and regulation: a debate over antitrust exemptions and government oversight

In a two-part series on the future of business, Decoder interviewed Jonathan Kanter, former head of the DOJ Antitrust Division, now a professor at WashU and Carnegie Mellon. Kanter discussed the growing debate over AI safety and regulation, with major tech companies like Anthropic, Google DeepMind, and OpenAI calling for antitrust exemptions to coordinate on safety measures. Researchers have raised concerns that AI could pose a greater than 10% risk of global extinction, prompting calls for slowed development and stronger regulation. Kanter argued that while companies may seek to collaborate on safety, this does not necessarily constitute a cartel. He emphasized that companies should be held accountable for building safe products, with potential liability for harmful AI actions. Kanter also highlighted the need for government intervention to establish safety standards, criticizing Congress for inaction on critical regulatory frameworks. Kanter contrasted the current regulatory vacuum with historical examples like automotive safety, stressing that innovation should not excuse companies from responsibility. He suggested that product liability laws could serve as a regulatory framework, citing recent cases against Meta for child safety issues. Despite calls for antitrust exemptions, Kanter maintained that collaboration on safety does not require such exemptions, as antitrust laws do not prohibit legitimate safety partnerships. The discussion underscored the complex interplay between innovation, regulation, and corporate responsibility in the AI landscape.