Join us in returning to NYC on June 5th to collaborate with executive leaders in exploring comprehensive methods for auditing AI models regarding bias, performance, and ethical compliance across diverse organizations. Find out how you can attend here.
The AI industry is undergoing a significant shift.
While the number of AI image and video generation tools continues to proliferate — just in the last two weeks, Google, OpenAI, and startup Krea have announced new ones — artists and photographers have also increasingly spoken out against their work being used to train AI models without compensation or permission. Some artists have filed class-action lawsuits against AI image and video generation companies such as Stability AI and Midjourney.
Now a new platform for creators, Wirestock, has emerged with a plan to hopefully satisfy both opposing sides: the platform allows photographers and artists to upload their works online and license them out through popular image services including Getty, Adobe Stock, and Canva, while at the same time, giving them the option to allow AI companies to train on that work in exchange for payment.
“All major [AI] players are quickly shifting to using ethically licensed content,” said Wirestock CEO Mikayel Khachatryan in a video call with VentureBeat last week. “This is partly due to legal pressures but also because it’s a practical solution for companies needing reliable data.”
VB Event
The AI Impact Tour: The AI Audit
Join us as we return to NYC on June 5th to engage with top executive leaders, delving into strategies for auditing AI models to ensure fairness, optimal performance, and ethical compliance across diverse organizations. Secure your attendance for this exclusive invite-only event.
Request an invite
Payment for artists and photographers for licensing their work from AI training ranges from microtransactions of cents on the dollar when providers buy bulk data from Wirestock, up to $15-20 per artwork for specific needs, according to Khachatryan.
“it’s a very, very lucrative source of revenue for artists, and they love doing it,” he adde.
Origin story: streamlining payment for creatives
Wirestock was founded in 2019 by Khachatryan and his co-founders Ashot Mnatsakanyan, Vladimir Khoetsyan, and Hovhannes Kuloghlyan.
The goal was initially to be a one-stop shop clearinghouse for photographers to publish work to multiple stock image services online.
“We built Wirestock as an aggregator platform to sell content on the biggest marketplaces like Shutterstock, Getty Images, Adobe, and Canva,” Khachatryan explained to VentureBeat: “Our goal was to simplify the process for artists and ensure they receive fair compensation.”
The idea was a simple one but with obvious merit: instead of each artist or photographer having to go to each marketplace and upload their work there, complying with that marketplace’s terms and requirements, Wirestock allows them to do the process once and automatically takes care of tagging the work appropriately for given marketplaces — and sorting which ones are accepting which types of work.
It also directly accepts the payment from the artists and provides them with their earnings, taking a small amount for its own revenue (15% commission on sales distributed to partner agencies and 30% on any direct sales).
AI presents a new opportunity
But as the generative AI industry took off, the founders recognized the opportunity to serve both the needs of AI model makers and their creative community of more than 500,000 users from 140 countries around the world.
Khachatryan explains, “Companies often struggle to find the exact data needed for their AI projects. Wirestock solves that by sourcing any type of visual content from its diverse community of creators at a large scale.”
Wirestock simplifies the workflow for Gen AI companies by offering a vast collection of ready-to-use, high-quality visual content that’s already been vetted and tagged by Wirestock with appropriate text descriptions and metadata — something that is necessary for the type of machine learning that AI model makers use to train their models what specific colors, image features, and objects look like so the model can generate them when prompted.
Wirestock’s approach to metadata includes atuomated keyword and caption generation using AI, which streamlined the content submission process.
Which AI companies are paying Wirestock for ethical human generated art and photography? Khachatryan declined to name specifics citing confidentiality deals, but said that the “the biggest companies developing models” including so-called foundation models — those trained from the ground up — were among its customers.
Out of Wirestock’s total community of creators, a majority — some 300,000 people — have already been paid through AI data set licensing deals, according to Wirestock’s CEO.
Creative challenges
In addition, Wirestock hosts time-sensitive “creative challenges” and paid projects for its users in the creative community — further diversifying their income streams. The challenges offer paid cash prizes for artists who create and upload work that fit a specific description of a Wirestock AI model customer or other enterprise’s needs.
For example, the current challenge at the time of this writing is around “Space Exploration” which asks artists to submit “awe-inspiring depictions of space” and “captures the essence of space exploration through your artistic lens.” The top three image entries judged by Wirestock and/or its customer will receive $25, $15, and $10 in prize money.
There are a number of requirements and suggestions for what types of images will be accepted and considered award-winning, including “Your submissions must be realistic, with good textures, and must not have incorrect proportions and incoherent colors.”
Khachatryan shared that “artists love participating in these projects as it provides a great way to monetize their work.”
AI training on AI generated art?
Intriguingly, like other stock marketplaces, Wirestock allows creators to upload AI generated works to its marketplace, which some have criticized as a form of “copyright laundering,” since the training data for those AI models may have included copyrighted works.
Wirestock says it attempts to label that content as AI-generated and asks its creative users who upload content to identify it as such, but ultimately leaves it in their hands as to where the content is created and by what means.
“We ask our artists to upload content using models that have made that commitment that their data is ethically trained,” Khachatryan said.
Yet Wirestock does not stop creatives from submitting work generated by Midjourney, Stability AI, or other models that scraped the web en masse, including potentially copyrighted works.
Where Wirestock goes next
Khachatryan also highlighted the growing demand for video content to train emerging AI video generation models such as, hypothetically, OpenAI’s Sora, which presents new revenue opportunities for visual creators.
“AI companies are looking to license video clips, which is becoming a substantial part of our business,” he said.
Through its stated commitment to ethical AI training and fair artist compensation, Wirestock seeks to be the leader of ethical AI data. Clearly, already many creators and artists agree and have decided to use the company to get paid for their work — one way or another.
VB Daily
Stay in the know! Get the latest news in your inbox daily
By subscribing, you agree to VentureBeat’s Terms of Service.
Thanks for subscribing. Check out more VB newsletters here.
An error occured.
{Categories} _Category: Applications,*ALL*{/Categories}
{URL}https://venturebeat.com/ai/wirestock-lets-photographers-and-artists-get-paid-when-ai-companies-train-on-their-work/{/URL}
{Author}Carl Franzen{/Author}
{Image}https://venturebeat.com/wp-content/uploads/2024/05/cfr0z3n_photo_of_three_diverse_attractive_adults_gathered_aroun_0f5bc4ad-9e42-4565-84fb-c5759324c434.png?w=1200&strip=all{/Image}
{Keywords}AI,Business,AI art,ai art generators,ai artwork,AI image editing,ai image generation,ai image generator,ai imagery,AI, ML and Deep Learning,console gaming,Conversational AI,diffusion models{/Keywords}
{Source}All{/Source}
{Thumb}{/Thumb}