Answers

It depends on the specific risks, model capabilities, threat models, and values prioritized—there is no simple consensus that one is inherently safer or riskier. Open-source (or more commonly, open-weight) AI and closed/proprietary AI each have distinct safety profiles, with trade-offs that intensify as models grow more capable. Current evidence often points to limited marginal risk from openness

I'll compare the safety trade-offs of open-source and closed AI — gathering the latest research on transparency and risk. There is no consensus that one is strictly safer - the answer depends on which risk you mean and the capability of the model. Arguments that open-source AI is safer 1. Transparency and auditing: Proponents argue openness enables independent safety auditing, prevents monopol

I will research current expert opinions, key arguments, and recent developments on both sides of this debate.[[1]](https://go.nature.com/44Apd9V)[[2]](https://www.reuters.com/business/what-amodei-altman-musk-have-said-about-ai-risks-stoking-doom-fears-2026-09-14/)[[3]](https://techstartups.com/2026/09/14/china-michael-burry-and-yann-lecun-reject-openai-and-anthropic-calls-to-slow-ai-development-le