Gies College of Business recently hosted a pioneering discussion about how consumers can reclaim ownership of their data and benefit from AI models built upon it.

The session, featuring Anna Kazlauskas, Open Data Labs CEO and creator of Vana, was organized by Vishal Sachdev, clinical associate professor of business administration at Gies Business. Faculty and students from various disciplines explored the revolutionary potential of decentralized data systems and their implications for the future of AI development.

The ‘data wall’ challenge

Kazlauskas explained that current AI models are predominately trained on publicly available internet data, which represents only 0.1% to 4% of all digital information. The rest sits behind  ‘data walls’, creating a significant bottleneck.

"We're actually running out of data to train AI models on," said Kazlauskas. "The public internet contains around 15 trillion tokens, and leading models have already been trained on that much data. To make them better, we need access to the remaining 96% of data that's currently trapped in walled gardens."

User-ownership explained

Vana’s mission is to build an infrastructure for individuals and companies to capitalize on user-owned data.

"What Vana fundamentally does is unlock data from these walled gardens and put it in users’ control,” said Kazlauskas. “Just as parking your car in a lot doesn't transfer ownership to the lot operator, storing your data on a platform doesn't surrender your legal rights to that information. Users legally own their data and can reclaim it from any platform through data export requests—a right protected by regulations like GDPR in Europe and CCPA in California.”

Kazlauskas noted that consumers can request valuable information about themselves from their digital footprint on platforms and companies such as Alexa conversations, Instagram advertising profiles, and Tesla driving records.

“I do research on AI and blockchain-based systems, and this is especially relevant to how data is priced and how users could be compensated for sharing it,” said Dmitri Sumkin, a postdoctoral researcher at Gies Business who teaches social media strategy. “It showed me how I can more directly connect data analytics and data labeling to real-world topics like data privacy and Web3.”

DataDAOs: Collective data governance

At the heart of Vana's vision are DataDAOs (Data Decentralized Autonomous Organizations). These user-governed pools allow individuals to contribute similar data types while maintaining control over its usage through collective voting mechanisms. This collective bargaining power enables users to dictate how their data is used and by whom, creating a powerful new dynamic in the data economy.

Vana currently supports about 50 active DataDAOs, which implement "proof of contribution" systems to ensure data quality and authenticity while preventing manipulation by bad actors. Notable examples include:

• Reddit DataDAO: 140,000 users contributing social media data
• Tesla DataDAO: Aggregating driving data for automotive and insurance companies
• Spotify DataDAO: Music streaming data for AI applications

“Vana’s approach enables cross-referencing of different types of data—for example, linking someone's fashion purchase history with their car purchase data—which opens up new possibilities for innovation through what might be called a 'cross-data' platform,” said Fei Du, associate professor of accountancy at Gies.

Privacy and Security Architecture

Addressing privacy concerns, Kazlauskas detailed Vana's technical safeguards. User data is encrypted with individual keys, then re-encrypted with DataDAO-controlled keys. Operations on this data occur within Trusted Execution Environments (TEEs), ensuring that raw data never leaves secure computing environments while still enabling valuable insights and model training.

"Only if the DataDAO votes ‘yes’ can the data actually be decrypted, and then only the specific operation that's been approved runs on that data in a secure environment," said Kazlauskas.  

Implications for Data Rights and Market Dynamics

The presentation sparked discussion about the broader implications of user-controlled data systems.

"Responsible AI begins with responsible data, and as a researcher in this space, I'd like to see innovative business models that prioritize user agency over personal data flourish and become more mainstream over time,” said Aravinda Garimella, assistant professor of business administration at Gies. “Thinking of data rights as labor rights is a powerful reframing that can spark important conversations about data ownership, collective bargaining, and algorithmic accountability."

Professor Ron Guymon highlighted the shifting value dynamics in the data economy.

"It's fascinating to observe how the relative market value of data, analytic skills, and business acumen changes as technology develops,” said Ronald Guymon, Gies senior lecturer of accountancy. “Given the current generation of AI tools for analyzing data, it makes sense that a data repository that provides a more holistic picture of behavior, along with business acumen, is more important than ever."

Gies College of Business continues to host cutting-edge discussions at the intersection of technology, ethics, and commerce. This session exemplified our commitment to preparing students for the evolving digital economy while addressing critical questions about data rights and AI governance.