Academic Paper: Anonymous-by-Construction. An LLM-Driven Framework for Privacy-Preserving Text
At Veritran, we turn research into concrete solutions.
Accepted by the world's largest international software engineering conference, the paper demonstrates how we apply artificial intelligence to solve one of the main challenges of enterprise adoption: leveraging data value without compromising privacy.
The research presents an anonymization approach that protects sensitive information while preserving context and data utility. A concrete example of how it’s possible to enable new use cases, products, and business opportunities, even in environments where trust and regulatory compliance are critical.
Privacy without Losing Value
The framework uses language models to identify sensitive information and replace it with fictional alternatives of the same type, preserving the context, structure, and usefulness of the data.
Sensitive Data Under Control
Information is processed locally, within the organization's infrastructure, before reaching external models, agents, or services. In this way, original data remains within the organization.
Useful Data for AI and Business
It is assessed that the anonymized information maintains its meaning, can be used by AI agents, and remains suitable for training models. This way, organizations can develop products, automate processes, analyze conversations, and transform services and operations without privacy and confidentiality requirements hindering innovation.
