Key Takeaway
Thales offers trial versions of AI security tools on its website, employing over 83,000 people globally in sectors like defense and cybersecurity. The company addresses security challenges in enterprise AI deployments using retrieval-augmented generation (RAG), which allows models to access up-to-date information. However, RAG can expose sensitive data. Thales’ RAG security capability scans data before ingestion, applying encryption and access controls to prevent unauthorized access to confidential information. This approach enables businesses to adopt AI confidently while protecting sensitive data and user interactions, leveraging Thales’ extensive security expertise.
Trial versions of various tools are currently available on the Thales AI Security Fabric website. The company employs over 83,000 individuals across 68 countries, specializing in defence, aerospace, and cyber security.
Thales tackles insecure RAG pipelines in enterprise AI deployments
Retrieval-augmented generation has emerged as a prevalent approach in enterprise AI deployments, enabling models to access up-to-date information without the need for retraining. However, RAG architectures present security challenges. Data retrieved from enterprise systems may include sensitive information that models could inadvertently disclose through their outputs.
Thales’ RAG security capability scans data prior to ingestion, implementing encryption and access controls to minimize exposure. The system is designed to prevent situations where confidential data enters model contexts and later appears in responses to unauthorized users.
“With decades of security expertise, Thales empowers businesses to confidently expand their AI adoption while protecting sensitive data, applications, and user interactions,” Sebastien states.
Leave a Comment