Cybersecurity experts at the RSA Conference this week will be on the hunt for ways to secure their operations in the era of generative AI.
They’ll find many of the latest tools use AI and accelerated computing. This intersection of security and AI is coming into focus with collaborations that companies like NVIDIA and its partners will describe at the event.
Data Science for a Data Problem
Machine learning is a great tool for cybersecurity because data is exploding.
“With more devices and users expanding the landscape to defend, cybersecurity has become a data problem; and AI is a data solution,” said David Reber, NVIDIA’s chief security officer.
Today, security analysts can be overwhelmed by a data tsunami. Generative AI can provide security copilots that act like filters, extracting from the firehose flow of information the context and anomalies that need a human expert’s attention.
Generative AI also lets security managers interrogate their data directly instead of setting rules, chasing alerts and tracking dashboards. In the age of AI, security experts will move from a command line to a conversational interface.
AI Puts Security Context on Steroids
This shift takes context-based security to a new level, according to a talk Reber delivered at GTC.
The potential is huge, but it requires work to unearth it. At GTC, Reber encouraged cybersecurity experts to begin working with AI, starting with low-risk use cases to identify and secure gaps.
He also provided suggestions for how to go about securing machine learning processes, saying security experts need to:
- secure data supply chains,
- develop tests for securing AI models and datasets across the development lifecycle,
- use model cards, data cards and software bills of materials to provide AI transparency and reliability,
- participate in community testing events such as security hackathons, and
- update policies on how to respond to AI security events.
Foundations for AI Cybersecurity
To give users a leg up, NVIDIA provides NVIDIA Morpheus, a cybersecurity AI framework that filters and classifies large volumes of real-time data. Morpheus, part of the NVIDIA AI Enterprise software suite, lets developers build applications that can detect spear phishing, insider threats and more.
Users can employ Morpheus with NVIDIA NIM and NeMo Retriever, microservices from the NVIDIA API Catalog for rapidly deploying AI. The combination can unlock new use cases, such as reducing from days to seconds the time to find and resolve common software vulnerabilities and exposures, one of many NVIDIA AI workflows.
A new release of NVIDIA DOCA — the software framework for programming NVIDIA BlueField DPUs and NVIDIA ConnectX NICs — provides another foundation for AI security. It now sports updated encryption features for network and storage data.
An Expanding AI Security Ecosystem
At RSA, many companies will show products built on NVIDIA technologies that extend security for the generative AI era, including:
- AIC will demonstrate Qrypt’s key generation for quantum secure encryption running on a BlueField DPU in an AIC-designed server.
- Anjuna will discuss how the U.S. Navy is evaluating confidential computing on the Anjuna Seaglass platform with proprietary LLMs running on NVIDIA H100 Tensor Core GPUs.
- Bloombase will show an updated version of its StoreSafe Storage Firewall powered by Morpheus and BlueField DPUs and new use cases for threat detection and fast, quantum-safe encryption of AI models and data.
- Check Point Software will show its AI Cloud Protect security solution on BlueField DPUs, Quantum Force security gateways on ConnectX NICs, and Quantum Maestro software on NVIDIA Spectrum switches.
- Cisco will feature Cisco Hypershield, an AI-native security architecture, to protect against both known and unknown attacks. It will also discuss its expanding partnership with NVIDIA to help customers harness the power of AI.
- CrowdStrike will show its CrowdStrike Falcon Foundry and CrowdStrike Falcon platform that employs NVIDIA’s GPU-optimized AI software, including NIM microservices.
- Deloitte will showcase CyberSphere, a cyber operations platform that uses Morpheus to speed detection of cyber threats.
- Palo Alto Networks will describe its collaboration with NVIDIA on two use cases, a next-generation reference architecture for securing generative AI deployments with NIM and its VM-Series Virtual Next-Generation Firewall, with expanded intelligent traffic offload (ITO), supporting BlueField-3 DPUs.
- Qrypt will demonstrate its quantum-secure key generation for securing AI workloads running on BlueField-3 DPUs using DOCA.
- Sygnia will announce the use of BlueField DPUs and Morpheus in Velocity Edge, its new hardware-accelerated MXDR service for the energy and industrial sectors.
They are part of the NVIDIA ecosystem building a new layer of security for generative AI. That community includes more than 20 companies at RSA this week from NVIDIA Inception, a program for cutting-edge startups.
At RSA, Daniel Rohrer, vice president of software product security at NVIDIA, will be part of the keynote panel on AI safety.
In addition, Kevin Deierling, senior vice president of networking at NVIDIA, will share insights on security at the Cloudflare Executive Supper Club. And NVIDIA will participate in an event about women in cybersecurity.
To get started with AI-powered cybersecurity, try a workflow in NVIDIA LaunchPad.