Trump control for US-style regulation

Trump has signed an executive order instructing federal agencies to test AI models 30 days before companies officially release them. This reflects the US approach to AI regulation through collaboration with tech developers. The aim is to allow the government to conduct cybersecurity testing before public release. This order, however, marks a shift from Trump's earlier preference for a more hands-off approach to AI oversight, as concerns grow about maintaining the US' lead in the field. China has adopted a more nuanced regulatory framework that combines ideological control with state support for innovation. The EU, meanwhile, employs a risk-based system that classifies AI technologies according to potential harms and imposes corresponding mitigation requirements.

Trump's executive order on voluntary testing may strike a workable balance. The US remains the world's leading AI innovation hub. Anthropic's Claude Mythos is being circulated among select government agencies before release because of concerns about their ability to exploit software vulnerabilities. The need for testing is not in dispute. But the debate centres on how it should be conducted by the Trump regime. Given that many AI-related vulnerabilities are shared across markets, achieving regulatory harmonisation is likely to take time.

AI regulation tends to reflect the structure of the underlying tech market. In the US, private investors drive innovation and are wary of government intervention. China relies more heavily on state-backed development and can afford to place less weight on investor concerns. The EU is primarily an AI consumer market. Its regulatory framework prioritises consumer protection. Over time, however, most AI markets are likely to become increasingly consumer-driven, shifting attention toward safeguarding people and organisations. A broader range of products and applications will be needed to generate market-led forms of AI regulation. Yet, the competitive race to establish AI leadership may be decided well before a mature mass market emerges.