Decentralized AI QA rising
New decentralized verification networks — Ambient L1, Nexus and Nous Research — are being discussed as programmatic ways to verify AI model quality and ‘slash’ underperforming providers, a setup that could automate fraud‑model governance. These systems are also being pitched for on‑chain commerce and card‑policy engines to flag risky AI behavior before it hits production Ambient L1/Nexus/Nous Research discussion.
Nous Research closed a $50 million Series A led almost entirely by Paradigm on April 25, 2025, a round that multiple outlets reported valued the project at roughly a $1 billion token valuation. cointelegraph.com Nexus announced the launch of its Verifiable AI Lab on May 8, 2025 and published a technical post on May 14, 2025 outlining an Incrementally Verifiable Computation (IVC) approach and a zkVM-based “Execution Layer” to produce succinct proofs of off‑chain computation. blog.nexus.xyz Ambient surfaced from stealth at the end of March 2025 with a reported $7.2 million seed round led by a16z’s CSX and public statements from co‑founders describing a proof‑of‑work L1 that intends to run AI inference on‑chain while remaining SVM/compatibility focused. techstartups.com Nous has published its “Psyche” architecture for Solana‑coordinated distributed training and announced strategic partnerships such as a May 2025 collaboration with Lambda to supply infrastructure for decentralized training, while Nexus has signaled a move toward large‑scale testnets and device‑to‑cloud compute claims in its product blog. nousresearch.com
