AWS re:Invent 2025 - Securing the AI Era: From Hype to Real-World Defense (AIM106)

Securing the AI Era: Zcaler's Approach to Real-World AI Defense

Transitioning to the AI Revolution

• The presentation discusses the shift towards intelligent systems, automation, and the growing prominence of AI and agents.
• The speaker highlights that the "AI revolution" is a "giga wave" that is subsuming previous "mega waves" like the internet, cloud, and zero trust.
• Zcaler's focus is on not just embedding AI in their products, but also extending their platform to secure AI for customers.

Evolving from Traditional Security Models

• Traditional security models based on trusted and untrusted networks, firewalls, and VPNs are no longer effective in the face of lateral movement and ransomware attacks.
• The speaker emphasizes that the network should not be the control channel for security, as IP addressing is a way to label traffic flow, not secure it.
• Zcaler's vision is to be a policy engine and switchboard between any source of traffic and any destination, providing fast, secure connectivity.

Implementing Zero Trust

• Zcaler has evolved their zero trust approach beyond just user access (ZTNA) to encompass third-party contractors, B2B users, workloads in the cloud, and physical sites.
• The goal is to secure communication across these diverse environments with a consistent policy framework, without relying on the network as the control channel.
• Key capabilities include:
  • Securing user access to applications without VPNs
  • Extending zero trust to third-party and B2B users
  • Securing workload-to-workload communication in the cloud
  • Securing physical sites and IoT devices with the zero trust branch

Securing Cloud Workloads

• Zcaler provides a comprehensive solution for securing cloud workloads across multiple cloud providers (AWS, Azure, GCP).
• Key use cases include:
  1. Securing workload access to the internet without the need for firewalls
  2. Enabling workload-to-workload segmentation using a consistent policy framework
  3. Supporting hybrid and multi-cloud scenarios with seamless connectivity and security
• Zcaler's cloud gateway service allows customers to secure cloud workloads without deploying any additional infrastructure.

Securing the AI Infrastructure

• The presentation highlights the growing importance of securing AI systems, including both public generative AI applications and internal AI infrastructure and applications.
• Zcaler has introduced AI Guard, a runtime guardrail service that inspects the communication between AI models and applications, leveraging small and large language models to detect potential issues.
• Zcaler has also acquired SPLX, an AI security company, to provide comprehensive visibility, governance, and red teaming capabilities for customers' AI infrastructure and applications.
• Key capabilities include:
  • Scanning and mapping the AI infrastructure and applications
  • Deploying pre-built adversarial attack simulations to test AI systems
  • Aligning with AI security frameworks like EU AI, NIST, and OWASP

Zcaler's Approach to Securing AI

• Zcaler emphasizes that they do not use customer data to train their AI models, and have a detailed policy on data privacy and usage.
• The company's goal is to provide a comprehensive platform that secures, simplifies, and transforms customers' digital environments, with a focus on securing data, applications, and workloads.

Key Takeaways

• Zcaler is taking a holistic approach to securing the AI era, addressing user access, cloud workloads, and AI infrastructure.
• The company's zero trust architecture and policy-driven security model are central to its strategy, enabling consistent security across diverse environments.
• Zcaler is investing heavily in AI-powered security capabilities, including runtime guardrails and comprehensive AI infrastructure security.
• The company's focus on simplifying and transforming customers' digital environments, while maintaining a strong security posture, is a key differentiator.