A journalist with expertise in film, television, and pop culture has transitioned from covering diverse topics such as comic books and social issues at io9 and Gizmodo to The Verge. Illustrators have long been vocal about generative AI companies that rely on artists' work for training their models without prior consent, arguing that such practices equate to theft. In response, advocates of generative AI have claimed that this approach is essential for advancing the technology. This ongoing conflict has sparked heated legal disputes and fueled the emergence of new AI startups, like Pippa, promoting themselves as more ethically responsible alternatives.
Pippa’s core offering, akin to its competitors, revolves around short, user-generated video clips that can be assembled to create visual narratives. However, what distinguishes Pippa is its commitment to compensating artists directly. Whenever a user generates content reflecting an artist's style, that artist receives payment from the platform.
Co-founders Hogan Shrum and Sean Wright believe that their dedication to partnering with artists will appeal to customers and demonstrate that ethical content creation is possible in the AI landscape. However, gaining the trust of artists presents a significant challenge. Despite its ethical aspirations, Pippa is still utilizing models that integrate the same sketchy practices seen in other text-to-video platforms.
During a recent conversation with Shrum and Wright, they emphasized Pippa’s mission to lead the industry away from the negative practices that have historically characterized AI development. Wright drew a parallel to the music industry, comparing the current situation to the days of Napster, suggesting that just as Apple found a way to ensure musicians were compensated, so too can AI companies find ethical methods for content creation.
While Pippa's remuneration for artists may not seem substantial—offering around $0.005 per image and $0.003 per video second—it proposes a system where increased user engagement boosts artists' earnings, supplemented by a 5% royalty pool based on the company's overall subscription revenue. However, this model has drawn scrutiny; the comparison to Spotify feels ironic given the ongoing disputes many artists have with the music platform over fair payment.
Launched in May, Pippa currently has about 800 paying subscribers and has entered into licensing agreements with four artists, with plans to bring on four more. Potential artist collaborators must first undergo a verification process to ensure the originality of their submitted illustrations. Upon approval, their art will be accessible to subscribers, with the added benefit of profile pages to promote their works beyond the platform. Notably, artists may choose to operate anonymously if there's concern about community backlash.
Despite its goal to transition to models trained exclusively on original datasets from their in-house artists, Pippa's technology still relies primarily on open models that utilize internet-sourced artwork without explicit creator consent. While Pippa aims for a constructive path forward, it struggles with the inherent challenges of creative ownership that plague generative AI.
This dilemma is not unique to Pippa; it's a broader issue for any startup claiming to offer "ethical" AI solutions, as they often use datasets not generated internally. Companies like Ben Affleck's InterPositive, on the other hand, emphasize creating completely original, proprietary datasets, although this approach requires significant investment that many startups lack.
Exploring Pippa’s platform reveals content primarily aimed at younger audiences, often mimicking Pixar’s distinctive style without capturing the nuances typical of independent artistry. This makes sense given that while Pippa seeks to showcase more original artistic styles, it continues to leverage many of the same cloud-based models as its rivals. Looking ahead, the incorporation of ByteDance’s Seedance 2.5 model may enhance its appeal by allowing users to create up to 30 seconds of customizable video without needing to restart the process. However, to truly differentiate from other platforms using the same model, Pippa must continue to attract more in-house artists, which hinges on its ability to effectively market the benefits to potential collaborators.
Shrum and Wright hold a hopeful outlook for artist recruitment, insisting that engaging with Pippa will encourage artists to promote the platform to their communities. This will require a fundamental shift in public perception regarding generative AI, ultimately favoring its acceptance and use.
