The ChatGPT Data Collective launched today, July 2, 2025, with the goal of giving users control over their interactions with OpenAI's ChatGPT, fundamentally changing the dynamics of AI training data ownership.

Users can participate by uploading their chat histories in a .zip file, which allows them to gain insights and earn $GPT tokens based on the quality and quantity of their contributions.

$GPT tokens serve dual purposes: they act as governance tools for decision-making regarding dataset access and usage, and they reward users for their contributions.

Governance of the collective is managed by a decentralized autonomous organization (DAO), promoting community-led decision-making and ensuring transparency regarding data usage.

Token holders have the ability to vote on various aspects of the shared dataset, including its usage, access permissions, and funding for tools.

This initiative is developed in collaboration with Vana, which emphasizes the importance of human ownership of data in the AI landscape.

The collective responds to ongoing criticism faced by large AI firms for using user data without consent or compensation, advocating for a privacy-centric, user-driven approach to AI training data.

User data remains encrypted until individuals choose to share it, and they have the option to update or delete their information at any time.

Anna Kazlauskas, founder of the Vana Foundation, highlights that the collective aims to restore control and ownership of data to users who have previously contributed to valuable AI datasets without their consent.

Future applications of the collective may include personalized journaling, memory analysis, and tailored AI experiences based on user data.

Participants in the collective are rewarded with $GPT tokens based on the quality and quantity of their data contributions, reinforcing the governance mechanism for the community.

The data submitted by users includes complete chat histories, AI comments, and metadata such as language settings and subscription levels, providing valuable insights into human-AI interaction.