• AIP Team

Private AI Solution Improved Customer Redaction of Private Information from 85% to 99%

USE CASE


Background


Keeping sensitive data private and secure has never been more important. Private AI was created to help companies do exactly this, by detecting and redacting personal, identifiable information (PII) through an automated AI model. Through their solution, Private AI makes it easy for businesses to weave privacy into their everyday workflows.


Contextual AI models utilize redaction, de-identification, and pseudonymization to strip over 30 categories of private information, including Personal Identifiable Information (PII), Protected Health Information (PHI) and Payment Card Industry (PCI), all without the need for third-party processing. With Private AI, it’s never been easier to keep your data safe, secure, and 100% yours.


Challenge


Private AI was approached by a client looking to stop PII leaking through their filters. An industry-leading provider of AI-powered speech-to-text transcription software, this customer kept noticing that sensitive information like credit card and social security numbers were bleeding into the final transcripts that their own clients were storing. This leakage put them and their customers at a significant risk should a data breach happen.


Unfortunately, the situation that this company found themselves in isn’t unique. Many businesses that work with personal data today struggle to keep PII secure within their systems. In order to help make better privacy a reality, Private AI worked on a solution using their AI-powered redaction tool.


Solution


To help their customer get their data situation under control, Private AI sat down with them to find out where these data leaks were happening, and how they could help plug the holes. Most businesses working with PII use regular expressions (regexes) to index and redact sensitive information. Regexes utilize a method of looking for patterns in text. Unfortunately this method isn’t the most compatible with real-world information, which can often come in a variety of different lengths, arrangements, and contexts.

Private AI used their contextual natural language processing (NLP) application to be able to understand the text more like humans can, in all its context. Through their NLP model, Private AI was able to help this organization produce cleaner transcript data and hit industry-leading accuracy in finding and redacting personal information. Ultimately, they improved customer redaction of PII in those call transcripts from 85% to 99%, giving their client more peace of mind about their data security.


“Private AI’s de-identification solution was extremely easy to integrate with our current pipeline, requiring only a few lines of code to ensure GDPR-compliant data handling for our users’ sensitive information. As a data anonymizer it was accurate and highly performant, allowing us to offer superior privacy protection without affecting our rates of service. Importantly, it enabled us to meet the rigorous data privacy requirements of the financial services sector without having to break the bank.” - Damian Tran, CTO Minerva AI


By leveraging the power of NLP over traditional redaction methods like regexes, Private AI could create a faster, more accurate, and more secure way to ensure businesses could keep their customer’s information private. In terms of speed, their solution can redact with 99% accuracy with a 45ms latency period. That’s 25 times faster than a reference NLP system on a single CPU core.


What’s more, they are the only fully AI-powered redaction tool that’s delivered to users as a Docker image. This means your data never leaves your environment, and is never shared with Private AI. And with the ability to understand information in 7 different languages, it’s a great tool for multinational organizations.


Conclusion


The state of global data privacy is on the decline. Every day, 30,000 sites are hacked, and 64% of businesses have reported experiencing a cyberattack at least once. Perhaps more concerning is this stat from March 2021, wherein 20 million records were breached in that month alone.


It’s easy to see why businesses need a trusted partner in protection more than ever. And with a solution that has outperformed every other redaction service currently on the market, Private AI is quickly becoming the preferred partner for these organizations.


“The docker image was really easy to integrate into our data workflows and we had it up and running in just a few hours. Our data involves mental health chat transcripts, so we were very happy to see that we were hitting impressive accuracy numbers out-of-the-box on a wide range of entity types that matter to our customers, saving us an enormous amount of time compared to building it ourselves. Plus their team was great to work with.” - Quinn Underwood, CEO Autumn AI

Contact


Don’t let a data leak get your business in trouble. If you want help keeping your sensitive data private and secure, contact us today to get connected with the Private AI team or check out the other use cases from our AI partners.

28 views