OpenAI Implements Refusal Feature in Response to Artist Concerns

OpenAI Implements Refusal Feature in Response to Artist Concerns

OpenAI has taken a significant step in the realm of artificial intelligence art generation by implementing a refusal feature in its model. This is most apparent in a feature that activates when users attempt to produce images that mimic the work of contemporary artists. Its goal is to address these mounting fears in the creative community. Beyond the letter of the law, this decision initiates a fascinating discussion regarding the ethical repercussions of AI-generated art. Most importantly, it highlights the need to protect original creators.

The new refusal feature as a matter of course protects the work of living artists. It refuses all demands to reproduce their one-of-a-kind looks. This initiative comes at a time when artists like Joyblivion, a 30-year-old artist from the Philippines, express their discontent with AI art trends. Since graduating from college, Joyblivion has been a full-time artist and she hasn’t looked back since. She sells most of her pieces through INPRNT, an online art-only marketplace. She has been [very vocal]( https://arxiv.org/pdf/2304.02433v3.pdf ) against the use of AI generators, calling the practice “so unethical.”

And in a recent appearance on content creator Varun Mayya’s podcast, OpenAI CEO Sam Altman vehemently justified what’s happening with AI art. That’s not the only reason he won over critics. He explained that though AI has made content creation easier and more accessible than ever before, it has created new obstacles for traditional artists.

“I think the democratization of creating content has been a big net win for society,” – Sam Altman

To his credit, Altman didn’t pull punches when it came to acknowledging many artists’ valid fears. Most of the art community feels like Joyblivion and remain actively engaged in tracking the advances of AI technology. Joyblivion has taken to social media platforms to share her artwork while condemning the use of AI and circulating videos that illustrate her viewpoints. One of those was by influential director Hayao Miyazaki, who called AI art “an affront to creation.”

Joyblivion reminded people who use AI to consider how it affects the livelihoods of artists.

“People who use it [AI generators] should be respectful of artists,” – Joyblivion (Joy Cardaño)

This great concern reflects the feelings of many in the artistic and creative community, including illustrators Sarah Andersen. Andersen is currently suing AI companies for using her work without permission. She underscored the need to give credit to artists and obtain permission before you start using their styles.

“Credit, consent and compensation, that’s what we would like from those models,” – Sarah Andersen

Andersen even pointed out the artistic impact, particularly for artists who make a living off of commissions. She observed that when you introduce AI, the ability for them to earn a living wage really goes down the drain.

“If you’re an artist that works mainly off of commissions, you’re probably losing a lot of income because it is cheaper, easier and faster to use a model to create an image than it is to pay an artist,” – Hollie Mengert

OpenAI’s efforts to mitigate these concerns have garnered attention, particularly after NBC News tested the model’s refusal feature. The investigation revealed that OpenAI’s system declined to generate images mimicking Andersen’s style as well as that of other artists, including Hollie Mengert.

Mengert further elaborated on the ethical and financial implications of AI-generated art, stating:

“And of course, that’s a financial issue, because if you want to use that particular artist’s style, you can do that, too, and you can do that without their consent, so it becomes a big financial issue.” – Hollie Mengert

The dialogue around AI and its implications for creative industries, particularly film making, is increasingly nuanced and dynamic. Many artists are still worried—very worried—about how technology could change the game for artists, and artistic creativity and expression. OpenAI has already built multimodal models like GPT-4o that create text, pictures, and voice. Yet, they are uniquely positioned to lead the way in prioritizing ethical innovation over innovative ethics.