Open models, despite their benefits, come with significant risks. Academics and businesses see them as tools for innovation and security. But AI safety proponents fear their ease of misuse by malicious actors. Governments and companies must weigh these concerns as they navigate a future where open models could democratise AI, or centralise power in dangerous hands.

The debate over open and closed models is complex and multifaceted. While open models like GLM-5.3 offer academic research and local business benefits, they can also be misused. Closed models, like those powering popular chat platforms, offer robust security but limit access to their inner workings. Decisions on which to support or limit will shape the future of AI, ensuring it serves the public good rather than a select few.

Dr David Gray Widder from the University of Texas at Austin highlights the influence of AI safety concerns on the debate. Those prioritising safety see open models as a risk, while those focused on practical applications see them as a way to spread AI's benefits more widely. Governments and companies must navigate these trade-offs, balancing safety with democratization to ensure AI's impact is positive for all.

Elizabeth Seger, from the Tony Blair Institute for Global Change, believes open models could prevent the mass concentration of AI power. She argues that these models challenge the control of AI infrastructure by a few powerful entities, ensuring AI remains a public good. The future of AI is not just about technological advancement but also about who gets to control its infrastructure and for what purpose.

As the debate continues, the outcome will shape the world we live in. A world where open models thrive could see AI's benefits spread more widely, or one where closed models dominate, leading to a more concentrated and potentially dangerous control over AI technology.