Robbie Barrat — open-source AI-art pioneer

2018

Artists: Robbie Barrat · Year: 2018 · Medium: AI-generated art and open-source neural-network code; early GAN experiments with classical-painting datasets · Substrate: StyleGAN, DCGAN training on classical nude-portrait and landscape datasets; code and trained models released on GitHub · Place: West Virginia / Stanford, USA · Institution: Stanford AI Lab (affiliate as teenager); independent Robbie Barrat — a self-taught teenage programmer from West Virginia — began publishing open-source GAN-based AI art in 2017 and spent the following year as one of the most productive and technically sophisticated contributors to the nascent field. Barrat was nineteen. He had taught himself deep learning through YouTube tutorials and the early Google Colab documentation. His GitHub repository — trained on art-history datasets he had assembled — contained pre-trained models that could generate work in the visual idioms of Rothko, Picasso, or the Old Masters, and Barrat released all of it under open-source licenses. The model weights Barrat released in 2017–18 became, nearly immediately, the specific technical infrastructure on which several early commercial AI-art projects were built. Most consequentially, the Obvious collective's Portrait of Edmond de Belamy (2018, elsewhere in this archive) — which sold at Christie's for $432,500 in October 2018 — was produced using Barrat's pre-trained GAN weights, a fact the Obvious collective did not initially disclose. The subsequent controversy was substantial: Barrat tweeted the source-code provenance publicly, Christie's and Obvious eventually acknowledged the derivation, and the incident became one of the foundational case studies of the ethical questions around AI-art authorship. What Barrat contributes to this archive is less any specific artwork than the critical infrastructure and the specific act of disclosure. He was one of the first AI-art practitioners to insist publicly that open-source model weights remained the intellectual labor of their original authors regardless of subsequent commercial deployment, and his willingness to call out the Edmond de Belamy provenance set a precedent the subsequent AI-art world has had to reckon with. Barrat has continued to work in AI-art and has produced personal NFT releases; he remains skeptical of much of the commercial AI-art market and has been an unusually candid public voice on its specific ethical failures. He is in his twenties now. Lore Robbie Barrat was nineteen years old in 2018, living in Palo Alto where he had moved alone from West Virginia for an internship at NVIDIA's research lab — he had finished high school early and had been entirely self-taught in deep learning through online resources. He uncovered the Obvious provenance issue from inside his NVIDIA cubicle by examining the specific GAN-output artifacts in the Christie's marketing imagery and recognizing them as outputs from his own published model weights. His decision to disclose publicly was not without internal conflict; several of his older mentors privately advised against picking the fight. Barrat has subsequently maintained an unusually candid public voice on AI-art ethics and continues to produce his own AI-collaborative work. He turned 26 in 2025 and lives in California; his earlier West Virginia upbringing is still legible in his public persona. Sources • Barrat, Robbie — GitHub repository (github.com/robbiebarrat). • Christie's — Portrait of Edmond de Belamy (auction record, Oct 2018). • Schneider, Tim — "Obvious Collective Faces Backlash" (Artnet News, Oct 2018).