Bria Secures Funding for AI Models - Licensed Data Focus

The Rise of Licensed Data in AI Image Generation

AI-powered image generators are currently facing legal challenges related to copyright, as numerous lawsuits have been filed against AI companies. These generators are often trained using extensive datasets gathered from publicly accessible websites.

Many companies defend their data collection and training methods by citing the fair use doctrine. However, this justification is contested by a significant number of copyright owners.

Bria: A New Approach to AI Training

Consequently, several startups and companies involved in the development of image generators are exploring an alternative strategy: training their models exclusively on content that has been legally licensed.

Bria, a New York and Tel Aviv-based company established in 2020 by Yair Adato and Gal Jacobi, exemplifies this approach.

Bria acquires images from approximately 20 partners, including Getty Images, and utilizes this licensed content to train its image-generating models, incorporating content safety measures.

According to Adato, Bria’s CEO, the platform provides “programmatic” compensation to image creators based on their “overall influence.”

Bria’s Capabilities and Features

“Bria foundation models contain one billion visuals and millions of videos,” Adato revealed to TechCrunch. “The company has successfully reduced biases that can sometimes appear in AI-generated visuals by training its models on diverse, globally representative datasets.”

“This ensures the models consistently produce visuals that reflect diversity, making them suitable for a wide range of creative applications.”

Bria provides plugins compatible with popular image editing and design software like Photoshop and Figma.

Additionally, a fine-tuning API is available, enabling customers to customize the company’s models for specific needs.

Users have the flexibility to run Bria’s models either on the company’s infrastructure or within their own computing environments, such as public cloud services.

In all scenarios, customers retain ownership of both the data and the resulting outputs, as stated by Adato.

“Enterprise clients can purchase access to the source code and [models],” Adato explained. “We offer over 30 specialized APIs for creating and modifying visuals, accessible through subscription and usage-based pricing.”

“Companies can invest in fine-tuning our generative AI models with their brand assets, creating bespoke engines that uphold their visual identity.”

Building an “IP Ecosystem”

Bria’s ambitions are substantial. Adato shared with TechCrunch that the 40-person company aims to establish an “IP ecosystem.”

This ecosystem will allow businesses to access licensed images from major media companies for use in commercial projects, ensuring “built-in compliance.”

The company also intends to broaden its platform and models to encompass additional media formats, including music, video, and text, as well as applications designed for on-device use.

“Bria is continuing to grow despite the challenges facing the broader tech industry,” Adato noted. “While the sector is experiencing headwinds from market saturation and macroeconomic pressures, these factors actually strengthen Bria’s position.”

Funding and Future Growth

While several companies, including Adobe, Spawning AI, and Shutterstock, are developing businesses centered around licensed media generators, Bria has successfully established a presence in this emerging market.

On Thursday, the company announced a $40 million Series B funding round, led by Red Dot Capital, with participation from Maor Investments, Entrée Capital, GFT Ventures, Intel Capital, and IN Venture.

This new funding brings Bria’s total raised capital to approximately $65 million, with the majority allocated to product development, according to Adato.

“We are experiencing rapid growth with our 40 customers, demonstrating significant annual recurring revenue growth exceeding 400% last year,” Adato stated.

“We are also expanding our team, adding expertise in key areas such as generative AI research and engineering for music and video, global sales and marketing, IP and copyright law, and generative AI consulting.

We anticipate doubling our team size by the end of the year.”

Updated 3/15 11:22 p.m. Pacific: A previous version of this article incorrectly stated Bria was founded in 2023, rather than 2020, and omitted the name of co-founder Gal Jacobi. We apologize for these errors.