Fortanix Confidential AI (C-AI) enables data teams, in regulated, privacy sensitive industries such as healthcare and financial services, to utilize private data for developing and deploying better AI models, using Confidential Computing.
The potential of AI and data analytics in augmenting business growth and advancement of solutions and services through data-driven innovation is well known. And that’s why AI adoption has been skyrocketing over the years. According to Gartner, by the end of 2024, 75% of enterprises will shift from piloting to operationalizing AI, driving a 5X increase in streaming data and analytics infrastructures. But MLOps often depend upon data that is sensitive and holds Personally Identifiable Information which is often subject to usage restrictions and carries compliance obligations. AI efforts can fail to move out of the lab if data teams are unable to use this sensitive data.
Realizing the full potential of AI is possible only if critical issues like data privacy and secure use of private information are addressed. And for most organizations within heavily regulated industries like healthcare and BFSI, large portions of such data still remains out of reach due to privacy concerns. For example: obtaining sufficiently large and meaningful clinical-trial data sets for training and predicting healthcare treatment outcomes is often challenging as its governed by privacy regulations like HIPAA.
Unlocking the full power of private data can only be done in a highly secure trusted execution environments. The emergence of Confidential Computing as a new security paradigm offers data scientists with a practical solution to the problem of protecting sensitive private data while being processed.
Fortanix Confidential-AI had been specifically designed keeping in mind the unique privacy and compliance requirements of regulated industries, as well as the need to protect the intellectual property of the AI models. Fortanix Confidential AI is offered as an easy to use and deploy, software and infrastructure subscription service.
Unlike traditional AI solutions that focus on accelerating modeling processes, Fortanix Confidential AI helps build richer models and protect the IP as well. It alleviates concerns of exposing private data by running datasets in secure enclaves and provides proof of execution in a trusted execution environment for compliance purposes.
Readily available and managed confidential infrastructure
Fortanix C-AI provides a readily available managed Confidential Computing infrastructure that’s easy to deploy and provision. The solution is powered on Intel Ice Lake third generation scalable Xeon processors.
Run AI models inside Intel SGX and other enclave technologies
AI models and frameworks are enabled to run inside confidential compute with no visibility for external entities into the algorithms.
Hardware-backed proof of execution and audit logging
The solution offers organizations with hardware backed proofs of execution of confidentiality and data provenance for audit and compliance. Fortanix also provides audit logs to easily verify compliance requirements to support data regulation policies such as GDPR.
Support for a broad set of AI/ML models
AI high performers have standard tool frameworks and development processes in place for developing AI models. Fortanix Confidential-AI supports a range of models such as Yolov 5, Decision Trees, SVM, Linear regression, KNN etc.
Protect data across all stages of MLops
The service provides multiple stages of the data pipeline for an AI project and secures each stage using confidential computing including data ingestion, learning, inference, and fine tuning.
Works with Azure
Fortanix managed and deployed infrastructure works on Azure AKS with SGX enabled nodes.
Dataset connectors support data ingestion
Dataset connectors help bring data from Amazon S3 accounts or allow upload of data from local machine.
Add security components like Tokenization and Encryption
Integration with Fortanix Data Security Manager allows organizations to further enhance security through additional components like key management and tokenization.
Fortanix is helping accelerate AI deployments in real world settings with its confidential computing technology. The validation and security of AI algorithms using patient medical and genomic data has long been a major concern in the healthcare arena, but it’s one that can be overcome thanks to the application of this next-generation technology
Glen Otero, Vice President of Scientific Computing at Translational Genomics Research Institute (TGen)
Some AI Project Use Cases
- Customer-service analytics
- Customer segmentation
- Targeted marketing
- Risk modeling and analytics
- Fraud and debt analytics
STRATEGY AND CORPORATE FINANCE
- Capital allocation
- M&A support
STRATEGY AND CORPORATE FINANCE
- Capital allocation
- M&A support
- Speech analytics and sentiment analysis to provide better tele-customer service in healthcare
- Fraud claim detection for health insurance
- Diagnosing medical conditions in radiology images
- Disease prediction using historical medical data
- Abnormalities and trend analysis within financial planning
- Capital market trends and mood analysis
- Fraud detection with ML
- Churn prediction for customers at risk
- Forecasting of claims and triage analytics
- Underwriting and other automatic workflows
- Purchase pattern analysis and targeted marketing
- Predictive purchase analysis using customer profile data
How Does the Solution Work?
Users can perform the following functions inside Confidential AI during the operational phase
- Provision Fortanix managed infrastructure for running curated applications as part of account creation
- Deploy pre-built curated AI models
- Connect datasets from external service and/or upload new datasets in a CSV format
- Create Confidential AI projects to create AI workflows
In Data ingestion phase, the data will be collected either by connecting to an S3 bucket or user can upload their data to the confidential AI platform.
In this stage, the users can choose the column names as a set of features and targets from the tabular dataset uploaded in the previous stage.
Building a model:
In this stage, the user has the option to build models by running a selection of pre-curated AI algorithms provided by Fortanix. The algorithms supported are: SVM (Support Vector Machines), Yolov5, Decision Trees, KNN, Linear Regression. Each algorithm supports a mode and data type for each mode.
In this stage, the data (images or tabular) is passed through a machine learning model to identify and predict the output from the data.