FLock.io (@flock_io): What a week in Washington, D.C.! 🇺🇸

What a week in Washington, D.C.! 🇺🇸 Our founder and CEO, @0x7SUN, joined @DCGco and 20+ founders across the decentralized AI ecosystem to propose how the Great American AI Act should approach decentralized AI. The Obernolte-Trahan bipartisan discussion draft is the most comprehensive federal AI framework proposed to date, seeking to establish a unified national standard for AI governance, enhance oversight of frontier developers, fund open-source security initiatives, and expand access to AI compute through NAIRR. All of these create meaningful tailwinds for decentralized AI. But just as early crypto regulation struggled to account for decentralized systems, the same challenge exists here. When AI training and inference are distributed across independently operated nodes through open protocols, and no single party controls development or possesses the complete model weights, there is often no identifiable "developer" to regulate, only contributors and node operators. As currently written, the Act's authorities could unintentionally extend to all of them. Our recommendations are straightforward: 1. Define a "decentralized AI system" as one where training or inference is distributed across independently operated nodes via open protocols, with no single person controlling development or holding the complete model weights. 2. Clarify that publishing open-source code or model weights, operating a node, or maintaining coordination protocols does not, by itself, make someone a developer or deployer under the Act. 3. Attach legal obligations to those with effective control over an AI system - the same principle FinCEN already applies by distinguishing software publication from money transmission based on control. This is not about creating a regulatory loophole. Anyone deploying AI in a regulated context should remain fully subject to the law. The goal is to ensure that liability follows control, not architecture. This matters because decentralized AI is no longer a niche blockchain narrative. It is already powering sovereign AI initiatives around the world. As the AI partner in UNDP's SDG Blockchain Accelerator Cohort 2, FLock has helped build sovereign AI capacity on decentralized infrastructure in the Dominican Republic, Liberia, and Rwanda, with similar work now underway with the state government of Sarawak, Malaysia. Governments across the Caribbean, Africa, and Southeast Asia are increasingly exploring decentralized AI for a key reason: sovereign AI capability without handing critical national infrastructure over to a handful of companies. Whether the AI serving that demand is built in America may come down to a safe harbor that protects and governs open, decentralized AI.