Get a Sneak Preview of Mozilla Data Collective’s Compensation Feature!

Mozilla Data Collective was built to redefine how AI data is created, shared, and governed. As part of our mission to be the data sharing platform for human agency and fair value exchange, we have long-teased what so many partners and community members have requested: a tangible way to ensure that data owners receive fair compensation from data downloaders in return for access to high-quality, consentful datasets. 

The wait is now over. We are so excited to announce the release of our compensation feature - starting today, we are making the first datasets available for paid download! Read on for more details on this sneak preview, how you can take advantage of the compensated datasets feature, and what’s next up in this exciting new chapter.  

Wayfinding in collaboration with Mozilla Data Collective partners

We have worked extensively with community partners to bring online some of the world’s most under-resourced language datasets, spanning Pashto (grassroots community), Catalan (Barcelona Supercomputing Centre, Project Aina), and Kinyarwanda (Digital Umuganda), to name only a few. Now, in response to the clear clarion call for equitable engagement and fair value returning back to these communities, we ran a couple of Calls for Proposals to source multicultural, multilingual, and multimodal datasets, and share them as a path toward a more sustainable model for ethical data sourcing. For each of these datasets, we engaged with trusted, close-to-context partners, and provided all requested compensation upfront in order to reduce the partner’s financial risktaking. This is just one tangible way in which we help to shift the market from exploitative gigwork and consentless scraping to one that centres human expertise and agency. 

Presenting three of our launch datasets

Brazilian Sign Language (LIBRAS) Healthcare Corpus. This large-scale multimodal corpus was developed by DataPopAlliance to advance inclusive AI and accessibility research for deaf communities in Brazil. It focuses on healthcare communication, including medical terminology, symptoms, preventive care, mental health, and clinical interactions, while maintaining an intersectional and gender-diverse approach.

This is an important gap to fill because the health domain remains deeply under-resourced outside English. There are too few high-quality, ethically sourced health datasets in other languages, and even fewer that are multimodal or designed around accessibility needs.

Medical Domain Mexican Spanish Speech Corpus (aka MEDMEX). This dataset includes 10.5 hours of medical-domain read speech from two registered nurses in our community, based in Puebla, Mexico. It was designed to support evaluation of Automatic Speech Recognition systems and training of domain-specific Text-to-Speech systems in Mexican Spanish.

Conversational Speech in Gujarati in the Medical Domain. This dataset was developed with our partner Karya and provides 25 hours of natural, multi-speaker conversational dialogue speech in Gujarati specifically recorded within the high-stakes medical domain. Gujarati is spoken by roughly 60 million people yet remains under-resourced in many AI systems, especially in critical domains across essential services, where language gaps can quickly become access gaps.

Visit Mozilla Data Collective and check these datasets out today!

How to take advantage of our compensation feature: Get in touch!

Together, these datasets point to the kind of ecosystem we want to help build: one where builders can access data that is useful, contextual, documented, and ethically sourced, and where the people who create those datasets are valued, recognised, and supported to find the right downloaders for their goals. 

For those creating, or seeking, high-quality datasets - drop us a note! We would love to hear from you. You can fill out our super-quick question form about your dataset offer or need, and our Partnerships and Data Curation teams will get back to you.

What’s next? Get ready for more, directly from data owners around the world

This sneak preview of our curation of multimodal, multicultural datasets is only the beginning. In very short order (we’re talking weeks, not months), the compensation feature will go LIVE for users, and our team will be supporting data owners to share their own compensated datasets that collectively drive toward a more multilingual, multicultural, and multimodal future grounded in human agency. 

If you want to see what an alternative, sovereign, and human-centric AI data economy can look like, sign up today and explore existing datasets. 

FAQs

What does this mean for data owners?

For data owners, this is a major unlock: the datasets they have built, curated, cleaned, documented, protected, and stewarded can now be shared with a clearer path to recognition and return. Whether the goal is to support research, sustain community-led data work, fund future collection, or, now, charge for access to high-quality datasets, data providers now have more ways to decide what participation looks like for them.

Who sets the pricing for datasets?

As a seller, you set the price for your dataset license. You may want to consider offering your dataset under different sets of terms and conditions for different audiences - for example, offering a dataset available under research terms and conditions for free, while listing it under commercial access terms for a price. You can do this by creating separate dataset listings for each version with the applicable terms for each, in conjunction with the individual access gate feature to review access requests before granting permissions. There is a $USD 100 base floor for datasets on the platform, and Mozilla Data Collective adds a low 5% platform fee to the transaction cost, which is paid by the downloader.

What are my primary responsibilities as a seller?

You are responsible for maintaining your datasets, ensuring legal compliance, setting licence terms, managing customer inquiries/refund requests, handling taxes, and maintaining your Stripe account. As a seller, you must provide a customer support phone number to Stripe, a dedicated support email address on your Mozilla Data Collective organisation page, and clearly communicate with buyers and/or MDC in the case of disputes.

How do I purchase a dataset?

To purchase a dataset, you will need to sign in to Mozilla Data Collective and navigate to the dataset listing. From there, you can view the terms and conditions for the license, and go through the checkout flow directly from the dataset listing page. The checkout flow is provided by Stripe, and will prompt you to add your payment details before you are able to download the dataset.