Thales launches AI Security Fabric to protect applications based on LLM and agent-based AI

A New Generation of AI Requires a New Generation of Security

Thales, a global leader in mission-critical cybersecurity solutions, has announced the release of the AI Security Fabric – the industry’s first comprehensive solution designed to secure applications based on Large Language Models (LLM) and autonomous Agentic AI. This platform is built to address the unique vulnerabilities that arise when using LLMs and the rapidly expanding capabilities of AI agents that can independently make decisions and perform actions.

Traditional security methods cannot cope with the dynamic and unpredictable nature of modern AI applications. Thales’ solution provides a unique set of controls for security and data privacy directly during AI model execution.

Key Capabilities of AI Security Fabric

Thales’ AI Security Fabric focuses on providing the control needed for safely deploying AI in the enterprise environment. The solution includes:

• Protection against LLM-specific attacks: Neutralizing threats such as prompt injection, data leakage during retrieval (Retrieval-Augmented Generation – RAG), and misuse of autonomous agents.
• Data access control and masking: Ensuring that LLMs have access only to the necessary data, while sensitive information is automatically masked before processing.
• Answer integrity verification: Using advanced techniques such as watermarking and cryptographic proofs to verify authenticity and origin of outputs, preventing harmful or false content generation.
• Usage policies and regulatory compliance: Allowing organizations to set strict policies on how AI agents may interact with corporate systems and data, ensuring regulatory compliance.

As AI reshapes business processes, organizations require security solutions tailored to the specific risks associated with Agentic AI and generative AI. Thales AI Security Fabric provides enterprises with specialized tools to protect AI applications while minimizing operational complexity. Leveraging its long-standing expertise in cybersecurity, Thales enables enterprises to confidently scale AI adoption while safeguarding sensitive data, applications, and user interactions.

Why Traditional Methods Don’t Work

The emergence of Agentic AI, capable of autonomously planning and performing multi-step tasks, creates fundamentally new attack vectors. Threat actors can manipulate agent logic or their interactions with external tools. Thales’ solution acts as a real-time “guardrail” that monitors LLM inputs (prompts), intermediate steps, and output data.

How Thales (Imperva) Can Help Protect AI

With AI Security Fabric, organizations can innovate with AI with confidence that their applications and sensitive data are protected. This platform is not a replacement for existing cybersecurity tools but represents a critically important, missing security layer specific to AI. It provides:

• Confidentiality: Ensuring that private or regulated data is not misused by AI.
• Integrity: Protecting AI application logic from manipulation and unauthorized actions.
• Availability: Preventing failures and abuse that could disrupt mission-critical AI services.

Thales (Imperva) offers a comprehensive approach by complementing existing application security solutions (WAF, API Security) with a specialized layer for controlling LLMs and Agentic AI.

Softprom Recommendation: Comprehensive Data Protection

In the context of protecting LLM applications, it is critical not only to control the AI itself but also to protect the foundational data it uses. This is where existing Thales technologies come into play, particularly solutions for encryption, key management, and database protection.

Thales (Imperva) technologies in the context of AI Security Fabric:

• Data-at-rest protection: Use the Thales Data Protection Platform to encrypt and tokenize data that LLM applications may retrieve from corporate storage. This ensures that even in case of a breach, the data remains unreadable.
• Data access protection: Privileged access management (PAM) and database monitoring (DBF) solutions from Imperva, now part of Thales Group, enforce strict control over which users, applications, or even AI agents access sensitive information.