Collaboratively authored by Anthony Cammarano, Mario Vargas, Muneeb Hasan, Alexandre Charlet, Andre Castro, Vic Levy, Ken Darker and Iwona Rajca
Generative AI (GenAI) applications are revolutionizing how businesses interact with data, primarily through Retrieval-Augmented Generation (RAG) pipelines, combining language models with vast enterprise knowledge bases. These pipelines allow organizations to query extensive internal datasets in real time. It is critical for organizations to do this to ensure that the right data privacy and security solutions are in place. AWS Bedrock has identified this need and offers native security guardrails within the Bedrock service.
Protegrity has a long history of of protecting sensitive data for organizations in regulated industries. Working with our customers, we have identified the need to provide enhanced guardrail capabilities to support their adoption of AWS Bedrock. As a result, we are pleased to announce the availability of the Protegrity GenAI Security for applications, an enhancement to AWS Bedrock’s native security framework.
Why Protegrity and AWS Bedrock?
Customers in regulated industries have unique data protection requirements. By leveraging Protegrity’s data privacy and compliance expertise, GenAI Security ensures that enterprises can meet their unique regulatory demands while fully utilizing their data within their GenAI applications.
Protegrity’s GenAI Security works seamlessly within the AWS Bedrock environment, extending the platform’s existing security controls to provide more comprehensive data protection.
Benefits for Enterprises Using AWS Bedrock and Protegrity GenAI Security
By integrating Protegrity’s GenAI Security with AWS Bedrock, organizations can confidently deploy their GenAI applications with the following benefits:
- Enhanced Privacy and Security: Sensitive data remains protected, even when accessed in real-time by LLMs, ensuring that only authorized users can see sensitive data within GenAI applications.
- Centralized Governance: Unified policies ensure consistency in data protection across GenAI applications, reducing the complexity of managing security at scale.
- Regulatory Compliance: Easily meet global data privacy regulations while maintaining data integrity and supporting audit and compliance requirements while adopting GenAI into your applications.
- Seamless AWS Integration: Protegrity GenAI Security complements AWS Bedrock’s security guardrails, enhancing data protection without disrupting workflows in your GenAI applications.
Customer example:
A multinational financial institution is deploying a GenAI-powered customer support application using an AWS Bedrock Retrieval-Augmented Generation (RAG) pipeline to provide real-time responses to complex customer queries. This application must handle vast amounts of unstructured data, such as conversational text, mortgage information, rate data, tax filings, and other potential sensitive data without clear governance and ownership. With the sensitive nature of the data involved, and proprietary financial information, the institution faces several significant challenges around data privacy and security with respect to this deployment.
Challenges:
- Sensitive Data Exposure: The generative AI application may need to access highly sensitive data (e.g., Social Security Numbers, proprietary rate schedules, tax IDs, etc), which could pose a significant risk if compromised or incorrectly exposed to unauthorized parties.
- Unstructured Data: Large amounts of data, including mortgage, rate, and tax documents, remain unstructured and loosely governed.
- Regulatory Compliance: Various regulations demand strict control over sensitive data with the ability to retrieve it when legally required (e.g., during audits).
- User-Specific Data: Certain data should only be visible to authorized individuals (e.g., customer service representatives vs. financial officers vs. end users).
- Content Moderation: Ensuring the safety and responsible use of an AI system that interacts with customers and sensitive data requires various methods for toxicity, hallucinations, and bias.
- Centralized Policy Management: The ability to enforce consistent policies across multiple AI models and data pipelines.
By augmenting AWS Bedrock’s existing guardrails with Protegrity’s GenAI Security, enterprises can confidently utilize their data within their GenAI applications and know that they are meeting the highest standards of data privacy and compliance. Whether you’re in finance, healthcare, or another highly regulated industry, GenAI Security offers the enhanced control and security needed to protect sensitive data, ensuring that the data used within your Generative AI applications remains safe and compliant.
Read up on our API Playground here, or try out a limited preview of our GenAI Security for yourself using our API Playground here.
See how Protegrity and AWS are creating safer and more effective AI-driven business solutions. Contact us to dive in.