Wirestock

Wirestock: licensed, creator-sourced training data for generative AI

Wirestock runs a two-sided marketplace between 700K+ creative professionals — photographers, videographers, illustrators, 3D artists, filmmakers — and the AI labs that need their output as training data. The company began as a stock-photo distribution service and completed a full pivot in 2025-26: today it supplies custom-structured image, video, illustration, vector, and 3D datasets for pretraining, SFT, RL, and RLHF workflows, and says six major foundation-model builders are customers.

For AI teams, the pitch is provenance and quality: data sourced from a consenting, paid creator network, structured to spec, and verified against rubrics — a contrast with scraped corpora and the legal exposure they carry. For creators, it is a way to earn from work as training data through submissions and paid challenges. A $23M Series A led by Nava Ventures closed in May 2026.

Key Features

• 700K+ creator network, 50M+ content pieces, 1M+ new pieces sourced monthly
• Custom multimodal datasets: image, video, illustration, vector, and 3D
• Supports pretraining, SFT, RL, and RLHF data workflows
• Dataset Library of pre-built sets plus custom commissioning
• Rubric-based quality verification and dataset structuring
• Creator Portal with paid challenges and tasks for contributors

Ideal Use Case

AI labs training generative image or video models that need licensed, rights-clean creative data at volume — especially teams whose counsel has ruled out scraping. On the supply side: professional creators with large portfolios who want recurring income from the AI training market rather than fighting it.

How Wirestock differentiates

General-purpose data vendors like Scale AI center on annotation labor; Wirestock's asset is the creator network itself, which produces original creative and generative-grade data rather than labeling someone else's. The pivot is recent — older reviews describe the legacy stock business, and its Premium creator subscriptions are being wound down — but the direction is funded ($23M, May 2026) and aligned with where model training is heading: licensed data with a clean chain of title.

FAQ

What is Wirestock? A marketplace where 700K+ creative professionals supply custom multimodal training data — images, video, 3D, illustration — to AI companies, following a 2025-26 pivot from stock-photo distribution.

How much does Wirestock cost? Dataset pricing for AI teams is custom and enterprise-negotiated; creators join free and earn per accepted submission or challenge.

Is the data licensed? Yes — data comes from Wirestock's own consenting, compensated creator network with rubric-based quality verification, rather than web scraping.

Who is Wirestock for? AI labs building image, video, or 3D generative models, and professional creators who want to monetize portfolios as training data.

tl;dr

Wirestock turned a 700K-creator stock community into a licensed training-data supply chain for AI labs — multimodal, rubric-verified, and rights-clean, with a fresh $23M Series A behind the pivot.