TIFIN has been granted U.S. Patent No. 12,705,394 for technology designed to protect sensitive personal information when working with machine learning models.
The patent, “Machine Learning Data Anonymizer,” covers technology that identifies sensitive information and replaces it with anonymized information before data is provided to an external machine learning model. The system is designed to preserve the context and meaning needed by the model without exposing the underlying sensitive information. Once an output is returned, the system can securely reintroduce the appropriate information for subsequent processing.
The work behind the patent began more than two years ago, as TIFIN was building AI-powered experiences for enterprises in financial services. The challenge was clear: how do you deliver deeply personalized experiences while operating within the privacy, security and enterprise-grade requirements of a regulated industry?
That challenge is particularly relevant to TIFIN atWork, where technology is used to guide employees through some of their most important financial moments. Doing that well requires personalization and context, while protecting the sensitive information behind those experiences.
Rather than accept the tradeoffs of an off-the-shelf solution, the team built its own.
The result was technology designed to anonymize sensitive information before it reaches an external machine learning model, while retaining enough context for the model to generate a useful response.
That same philosophy continues to shape how TIFIN builds today. Across TIFIN and platforms including TIFIN atWork, the focus is not simply on AI that demos well, but AI that can operate in production within regulated environments.
That means evaluating technology across quality, cost and user experience, while also accounting for privacy, security, latency, regulatory requirements, auditability and transparency.
Sometimes the right answer is an off-the-shelf component. And sometimes the requirements of the problem mean building something yourself.
U.S. Patent No. 12,705,394 recognizes one example of the latter.
Congratulations to Sarah Reid, the named inventor on the patent, and the broader AI team across TIFIN whose work continues to advance how we build and deploy AI for financial services.
TIFIN Media Contact
Tanya Bhasin
Tanya@tifin.com
408-332-1750