Calls to slow the rapid development of artificial intelligence (AI) systems have sparked debate over whether such a move is feasible amid intense global competition. Dario Amodei, CEO of Anthropic, a company that develops the AI system Claude, recently urged caution in advancing increasingly powerful AI technologies, citing concerns that safety measures and regulatory frameworks are not keeping up with the pace of innovation.

The challenge lies in the competitive environment surrounding AI development. Major technology firms including OpenAI, Anthropic, Google, and Meta are vying for market share, investment, talent, and technological dominance. This raises questions about whether companies can afford to impose self-restraint if doing so results in losing ground to rivals advancing rapidly.

Research indicates that the gap between open-weight AI models and leading closed proprietary systems can develop quickly. A study by Epoch AI found that since the start of 2026, open models have lagged behind the most capable closed models by an average of approximately four months on a capability index. While this gap does not apply uniformly to all models, it underscores the speed at which advancements are occurring and the limitations of relying solely on voluntary slowdowns.

Governments worldwide have introduced principles and guidelines addressing fairness, transparency, accountability, privacy, and safety in AI development. Malaysia, for example, has implemented National Guidelines on AI Governance and Ethics. Though these frameworks represent important progress, experts caution that establishing principles is distinct from enforcing them in practice, especially when difficult decisions about releasing powerful AI systems arise.

Key issues include determining who should evaluate AI models before release, the disclosure requirements for developers, and which entities hold decision-making authority over safety assessments. Equally significant is the risk that companies adhering to stringent safety protocols might be placed at a competitive disadvantage compared to others that prioritize speed over caution.

Additionally, the rapid pace of AI innovation presents regulatory challenges. AI capabilities can evolve considerably within a matter of months, often outpacing government efforts to formulate and implement effective regulations, which may take years. Moreover, overly complex or costly compliance mechanisms could inadvertently reinforce the dominance of large technology firms capable of bearing such burdens.

For Malaysia, which aims to establish itself as an AI nation by 2030, these considerations are particularly salient. While the country can influence how AI is applied within local universities, businesses, and public institutions, much of the most advanced AI development occurs abroad. Consequently, Malaysia’s AI strategy must extend beyond responsible adoption to include building expertise for evaluating AI systems, understanding their limitations, and maintaining appropriate levels of human oversight.

International collaboration will also be critical, given that no single country can manage the global ramifications of AI development alone. As Malaysia and others grapple with these challenges, the fundamental question remains whether current global dynamics allow for a meaningful slowdown in AI progress without compromising competitive interests.