{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"AI Security Ops","title":"Banning Open Weight Models | Episode 66","html":"","width":"100%","height":180,"duration":1956,"description":"In this episode of BHIS Presents: AI Security Ops, the team tackles a deceptively simple question with some very complicated answers:\nCan you actually ban an AI model?\nNot access to an API. Not the chips used to train it. The model weights themselves — files that can be downloaded, copied, modified, quantized, fine-tuned, and redistributed around the world.\nAs governments consider restrictions on Chinese open-weight models, the security argument cuts in both directions. There are legitimate concerns around national security, guardrails, model capabilities, and foreign technology dependence. But those same open models are inexpensive, locally deployable, and can give defenders capabilities that commercial frontier models sometimes restrict.\nSo what would a ban actually accomplish — and could it even be enforced?\nWe dig into:\n- Where U.S. restrictions on open-weight models currently stand\n- Why banning downloadable model weights is fundamentally different from restricting an API\n- How procurement rules and hosting restrictions could create a “soft ban”\n- Why the economics of open-weight models are driving adoption\n- How restrictions could disproportionately impact startups and smaller organizations\n- Whether modifying, quantizing, or fine-tuning weights makes model-specific bans impractical\n- The national-security argument for restricting Chinese models\n- Why guardrails on hosted frontier models matter to the security debate\n- How Hugging Face turned to a locally hosted open-weight model during incident response\n- Whether banning open weights could put defenders at a disadvantage\n- How existing government actions can indirectly limit access without banning a model outright\n- The hardware and operational costs of self-hosting large models\n- China, AI infrastructure, market competition, and industrial-scale distillation\n- Anthropic’s argument for mandatory safety testing of sufficiently capable models\n- Why safety testing gets complicated when open-weight guardrails...","thumbnail_url":"https://img.transistorcdn.com/mN9_Xu9UJwoaajIvIvLd-Yygv-Vh_nJwEDItjPY09kA/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS8zYjBm/MzE1MWI2YmE4ZGJh/MDQ3MmJkMTkxZGNl/MjBjNS5wbmc.webp","thumbnail_width":300,"thumbnail_height":300}