Pippa pays artists per AI image—but still uses scraped art underneath

Pippa's launch reflects a broader industry shift in response to sustained artist criticism of generative AI. For years, illustrators have pointed out that most text-to-image and text-to-video models were trained on their work without permission—a practice many AI boosters justified as necessary for technological progress. This gap between creator interests and industry norms has sparked legal disputes and given rise to startups explicitly marketing themselves as ethical alternatives.

Pippa's payment model attempts to bridge this gap by compensating artists directly whenever subscribers generate images or videos derived from their style. Cofounders Shrum and Wright frame this as analogous to Apple's iTunes store solving the Napster music-piracy crisis—a neat historical parallel, though one that glosses over Spotify's own controversies with artist compensation. The comparison suggests their theory: if the economics are sufficiently visible and fair, artists may overcome community stigma and join.

Yet the company faces an inherent structural tension. While Pippa markets itself as ethical, its current technology still incorporates art scraped from the internet without express consent, because the open-source models it relies on were trained that way. True originality—a proprietary dataset built entirely in-house—requires substantial upfront capital that most startups lack, and even then produces specialized tools rather than consumer-grade services. This means Pippa can improve incentives for partner artists, but it cannot cleanly solve the foundational problem it claims to address. Whether artists view this honest-but-incomplete approach as progress or as insufficient remains the central question for its growth.