Will the U.S. ban any open-source AI model in 2026?
As of 23:41 UTC on 7 Oct 2026, Kalshi prices “Will the U.S. ban any open-source AI model in 2026?” at 0%. The market has no Kalshi-reported volume in the last 24 hours. pdata.world compares it with 23 related events on Polymarket, Manifold and Kalshi.
If a federal ban on public access to open-source AI models is enacted in the United States through legislation, an executive order, a regulation or another binding federal measure before Jan 1, 2027, then the market resolves to Yes.
Enactment requires completion of all constitutional and legal requirements for the measure to become legally binding. Standard legislation requiring executive approval: final passage by all required legislative bodies, executive signature OR becoming law despite executive inaction OR successful veto override, and satisfaction of any required waiting periods Executive orders: official signature, publication in official government records, and satisfaction of any required procedural steps for implementation; a delayed effective date does not prevent enactment Administrative regulations, including export controls: completion of required rulemaking procedures and final publication in the official government record; a binding interim final rule counts once published Other governmental measures: official issuance in final, legally binding form The following do NOT constitute enactment: Legislative passage without completed executive action Passage by only some required bodies Provisional passage Non-binding resolutions Draft orders or proposed rules Enactment occurs when all requirements are completed, regardless of pending judicial challenges. A qualifying measure counts once enacted or issued, regardless of a later effective date. For this market, a “ban on open-source AI models” means a legally binding federal measure that prohibits distribution to, or access to or use by, the general public in the United States of at least one general-purpose large language or multimodal foundation model whose weights are publicly downloadable and usable on user-controlled hardware, such as Llama, Mistral, DeepSeek or Qwen. Source-available models qualify regardless of license restrictions on acceptable use. The measure may target one model, a named developer or country, or all such models, and may be temporary or indefinite, provided it bars ordinary public access through ordinary channels in the United States, even if copies remain available from foreign sources or prior downloads. A measure limited to government devices, federal contractors or procurement; a measure that bars only hosted apps, websites, chatbot or API services offering the model, or removes the model from only one app store, hosting provider or other single channel, while U.S. hosts may still lawfully distribute the model's weights to the public; an export control that restricts only transfers abroad; a state or local measure; a licensing, registration or reporting requirement that does not bar public access to an existing model; or a measure covering only narrow or special-purpose, deprecated, research-only or preview-only models does NOT qualify.
