Protegrity Introduces Open-Weights Model to Improve Detection of Sensitive Personal Data - Cybersecurity Insiders
Cybersecurity company Protegrity has introduced Protegrity-PII, an open-weights AI model designed to identify and classify 69 types of personally identifiable information across enterprise applications and AI workflows.
Available on Hugging Face, Protegrity-PII is a compact 278 million parameter model capable of detecting and classifying 69 types of personally identifiable information. The model covers sensitive data including account and card information, health records, credentials, device identifiers and biometric data.
Benchmark Performance
The company also released PII-SCOPE, a public benchmark with 8,030 records across healthcare, finance, insurance, retail and government sectors. Protegrity-PII achieved 0.910 typed recall and 0.902 precision on this benchmark, outperforming tools including AWS Comprehend, Microsoft Presidio and OpenAI Privacy Filter.
Our research represents a major PII detection breakthrough, one that advances our model capability far beyond the field, and we are scaling towards a model that is performant enough to be trusted by the world's most risk averse enterprises.
Technical Approach
PII detection sounds like a solved problem until you look closely at how existing approaches perform against the breadth and complexity of sensitive data found in real-world environments.
Built on XLM-RoBERTa, the model uses token-level classification to identify sensitive information, treating text as content rather than instructions. This approach reduces exposure to prompt injection attacks. The model can run directly within an organization's infrastructure, allowing security teams to inspect sensitive information across prompts, tool calls, APIs and agent responses.
Related
