Will compensating artists encourage them to adopt AI technology?

Illustrators have been raising concerns for years about generative AI startups using their artwork without consent to train their models. They argue that this practice amounts to theft, while proponents of generative AI claim it’s essential for the technology’s growth. This ongoing debate has sparked legal disputes and led to the emergence of new AI startups like Pippa, which positions itself as a more ethical alternative.

Pippa primarily offers short, user-generated video clips that can be combined to create visual narratives. What distinguishes Pippa from other generative AI video platforms is its commitment to compensating artists directly. Whenever a Pippa subscriber creates an image or video inspired by an artist’s style, that artist receives payment from the platform.

Co-founders Hogan Shrum and Sean Wright believe that their dedication to collaborating with artists will attract users. They aim to demonstrate that AI companies can ethically produce content. However, Pippa faces challenges in gaining the trust of artists. Despite its ethical claims, the company is still utilizing the same flawed models as other text-to-video services.

In a recent conversation, Shrum and Wright expressed their vision for Pippa as a way to move beyond the “bloody history” of the AI industry. They want to set a precedent for their peers, showing that AI can thrive without resorting to theft. Wright drew a parallel to the music industry, recalling how Napster facilitated music piracy until Apple introduced a legitimate model for compensating artists.

While Pippa’s payments to artists are modest—$0.005 per image and $0.003 per second of video—users pay between $14.99 and $99.99 for subscriptions. As users generate more content, artists’ earnings increase, and they also share in a 5 percent royalty pool funded by subscription revenue. Pippa compares its payment structure to that of music streaming services, a choice that raises eyebrows given the criticism many artists have directed at those platforms.

Since its launch in May, Pippa has attracted around 800 paying subscribers and has signed licensing agreements with four artists, with discussions underway to onboard four more. Artists must undergo a vetting process to ensure their submitted illustrations are original. Once approved, their art becomes available for Pippa’s subscribers. The platform also offers artists a profile page to promote their work outside of Pippa, and they can choose to use pseudonyms if they fear backlash from their communities.

Wright noted that many artists express enthusiasm for Pippa but worry about potential ostracism. He believes that as more artists see the financial benefits and understand the company’s intentions, participation will increase.

Pippa aims to transition to models trained exclusively on datasets from its in-house artists, but currently, its technology still relies on open models that have been trained on a broader range of content, some of which may have been sourced without consent. While the company’s goals are commendable, it still grapples with the fundamental issue of creative theft inherent in generative AI.

This challenge isn’t unique to Pippa; it applies to all “ethical” AI startups that utilize external data for their models. Some companies, like InterPositive, emphasize their development of entirely original datasets, but this approach requires significant upfront investment, often resulting in specialized tools not aimed at general consumers.

Exploring Pippa’s online portal, where users can view and create generative AI animations, I found the content largely similar to offerings from other text-to-video services. Much of it appeared to target children and struggled to replicate the polished aesthetics of major studios like Pixar, lacking the unique touch of independent artists.

While Pippa plans to integrate more original art styles, it still relies on many of the same cloud-based models available to other platforms. In the near future, Pippa intends to incorporate ByteDance’s Seedance 2.5 model, which could attract subscribers due to its ability to generate up to 30 seconds of video that can be refined without starting over. However, to truly differentiate itself from competitors using Seedance 2.5, Pippa will need to onboard more in-house artists, which hinges on its ability to effectively communicate its value to potential partners.

Shrum and Wright are optimistic about attracting more artists and believe that using Pippa will encourage them to promote the platform within their communities. For this to happen, though, public sentiment regarding the use of generative AI will need to shift back in favor of the technology.