Palo Alto Networks Chairman and CEO Nikesh Arora touted during the vendor's latest earnings call that it started its artificial intelligence (AI) journey seven years ago, and that it plans to deploy its own proprietary large language models (LLMs) in the coming years to capitalize on the ongoing “AI renaissance."

“There is a renaissance in artificial intelligence driven by significant advances in large language models,” Arora said.

Palo Alto Networks sees opportunities to incorporate LLM and generative AI into its products and workflows. Generative AI could improve core detection and prevention efficacy, facilitate more intuitive, natural language-driven product experiences and drive operational efficiencies across the enterprise.

“We intend to deploy proprietary Palo Alto Networks security LLM in the coming year and are actively pursuing multiple efforts to realize these three outcomes,” Arora said.

He added that the company plans to work with every public and open source model of generative AI to understand how it can build its own LLM using its proprietary data.

To put this into perspective, Palo Alto Networks analyzes nearly 750 million new telemetry objects globally in a typical day, including files, URLs, domains and domain name system (DNS) connections. And the company processes more than 3.5 petabytes of data per day.

Currently, the vendor uses about 1,000 AI models to analyze that data, detect threats and block 8.6 billion attacks across its customer base daily, Arora touted. “This all proprietary is happening. In our instance, this is not an LLM that's going out and getting trained, this is a proprietary AI model used by Palo Alto Networks, built by Palo Alto Networks being used for a specific use case and tasked for security,” the exec said.

Over the past seven years, Palo Alto Networks has gradually introduced machine learning (ML) capabilities into its offerings, and accelerated those efforts over the last two years, Arora stated.

The company began the journey by integrating ML models into its malware protection engine WildFire. In 2020, Palo Alto Networks launched what it calls the “world’s first” ML-powered next-generation firewall, with AI as the core of the firewall to stop threats, secure IoT devices and recommend security policies essentially in real-time.

Since then, the vendor has embedded AI capabilities across almost all of its security subscriptions, from DNS Security to advanced threat prevention, Arora noted.

This AI integration is also in line with the company’s push for autonomous security operation centers (SOCs) to replace legacy security information and event management (SIEM) and SOC tools.

Additionally, Palo Alto Networks baked AIOps into its secure access service edge (SASE) and SD-WAN services, operational technology (OT) and IoT security. The company claims it put AI and ML at the center of Its SASE plans for the upcoming year, according to SVP Kumar Ramachandran.

The increasing prevalence of technology vulnerabilities and the diminishing reaction time for security vendors and customers highlights the need for ML and AI tools to be used in SASE and security, Ramachandran told SDxCentral in an earlier interview.