Open-Source AI and Open Models Reading List
Posted on Hacker News, this item is a shared research bibliography from an author preparing public-audience and policy-facing writing on open models. The author says the list collects the best writing on open models from the last few years, aims to give anyone a comprehensive overview of the state of affairs if they want to get up to speed, invites commenters to suggest pieces to add, and plans to update it over time. The list was last updated 13 Sep. 2026.
Organized around what open models are, why people release them, how they relate to business strategy, and what the risks are, the entries include several clusters of work.
On strategy and economics, the list cites Bill Gurley's "From Open Source Software to Open Source Strategy" (May 2026) as a walkthrough of how open-source software has been used by businesses and early signs of what that means for AI; Mark Zuckerberg's July 2024 "Open Source AI is the Path Forward" as one of the clearest articulations of why Meta releases open models; Irene Solaiman's Feb. 2023 "The Gradient of Generative AI Release" on viewing release on a gradient rather than an open/closed binary, based on licenses, cost of running the model, data access, etc.; Nathan Lambert's March 2026 "What comes next with open models" on open models as a complement to strong closed models and for custom agentic workflows in enterprises worldwide; Christian Catalini's Aug. 2026 "Some Simple Economics of Open versus Closed AI" on how open models capture value as a complementary tool to large swaths of the existing economy, drawing on IP history and current debates such as distillation; Lambert's Feb. 2026 "Open models in perpetual catch-up" on why open models will constantly trail closed models in performance; and Lambert's June 2026 "Open and closed models are on different exponentials" on where adoption differs.
On safety, risk, and data, it includes Thinking Machines Lab's July 2026 "A Safe Path to Open Weights" on balancing powerful open-weight releases with safety; Kapoor, Bommasani et al.'s Feb. 2024 "On the Societal Impact of Open Foundation Models," an early paper on marginal risks that showed text-focused LLMs only very marginally increased documented potential risks of models; Florian Brand's June 2026 "The Myth of unsafe Open Source AI" arguing closed-model safety guardrails are regularly bypassed and cause many real AI risks before hypothetical open-weight risks have emerged; and Shayne Longpre et al.'s July 2024 "Consent in Crisis: The Rapid Decline of the AI Data Commons" on the mass reduction in open data that has hampered truly open AI research. The excerpt also begins to reference recent examples on how strong Chinese mod— (truncated).