Open models and safety pacts put control at the center of AI debates

Recent AI discussions have highlighted a question beyond whether the technology needs rules: who controls the models and how they are used. A TechCrunch Disrupt discussion is set to examine whether founders build on open models, which outsiders can adapt, or closed ones, where access remains under the provider’s control. That choice can affect both competition and safety: openness can give startups more freedom and reduce dependence on a few suppliers, but it can also make limits harder to enforce once a model is widely shared. Separately, calls for voluntary safety commitments show the limits of self-regulation: companies may endorse principles, but a nonbinding pact does not compel compliance.