Response to NTIA Request for Comments on Dual Use Foundation Artificial Intelligence Models with Widely

These risks include all of the following: ● The promotion of harmful stereotypes, violence, and online radicalization ● The potential manipulation or persuasion of large groups of individuals to take harmful actions, via the proliferation of chatbots without adequate safeguards ● The spread of non-consensual intimate imagery (NCII) including child sexual abuse material (CSAM) ● The spread of other. [...] How can making model weights widely available improve the safety, security, and trustworthiness of AI and the robustness of public preparedness against potential AI risks? The state of the field of AI as we know it has greatly benefited from the open release of data, models, and their weights. [...] That is, it could help 23 reduce the centralization of data and power in the hands of a few technology companies, but it would be non-trivial for the government to match the quality of leading private models and to learn how to monitor their usage effectively. [...] 14110 grants the Secretary of Commerce the capacity to adapt the threshold, is the amount of computational resources required to build a model, such as the cutoff of 1026 integer or floating-point operations used in the Executive Order, a useful metric for thresholds to mitigate risk in the long-term, particularly for risks associated with wide availability of model weights? One of the challenges. [...] 14110 grants the Secretary of Commerce the capacity to adapt the threshold, is the amount of computational resources required to build a model, such as the cutoff of 1026 integer or floating-point operations used in the Executive O...