Palo Alto expands capabilities to secure agentic AI
Is cybersecurity entering a phase in which it no longer only detects threats, but begins to remediate them autonomously?
Palo Alto Networks is betting that the answer lies in the new era of agentic artificial intelligence.
Palo Alto Networks introduced new capabilities aimed at securing agentic artificial intelligence environments, with the evolution of its Prisma AIRS platform to version 3.0 and the expansion of its secure browser, amid accelerating enterprise adoption of autonomous agents and rising risks associated with these systems, according to CEO Nikesh Arora.
“We are probably at the biggest technological inflection point of our lifetimes until the next one” Arora said during a session with specialized press, which included RadarTecnologico.com, alongside other international attendees.
Prisma AIRS 3.0: security for AI agents.
The company’s central announcement was the evolution of Prisma AIRS into version 3.0, designed to extend protection in environments where AI agents are increasingly being integrated into business processes.
According to the company, the new version introduces capabilities such as agent artifact analysis, automated security testing (red teaming), runtime controls, and a unified platform for managing the full agent lifecycle.
“The biggest shift is that it is no longer enough to find problems; we now have to fix them”, Arora said, describing the strategic move toward security models capable not only of identifying vulnerabilities, but also of automatically remediating them.
The company also introduced an agentic gateway architecture designed to centralize visibility over AI agent traffic within organizations, along with identity mechanisms tailored specifically for these systems, amid a growing expansion of agents across enterprise environments.
Secure browser and SMB expansion.
Another key announcement was the expansion of Prisma Access Browser, the company’s secure browser, which already has nearly 10 million contracted deployments. The tool aims to integrate artificial intelligence usage directly into the browser while maintaining security and governance controls over interactions.
Palo Alto Networks also announced a version targeted at small and medium-sized businesses (SMBs), with an annual subscription model and simplified deployment. The company positioned this version as a way to extend security to organizations with lower technological maturity, in an environment where browser-based AI applications are rapidly expanding.
Emerging risks: autonomous agents, identity, and supply chain.
During the presentation, Arora warned about the risks associated with deploying autonomous agents in enterprise environments, particularly in relation to credentials, excessive permissions, and attacks on the AI capability supply chain.
“If agents begin to act autonomously without supervision, the risk is that they will execute non-deterministic decisions that impact enterprise infrastructure”, the CEO said, referring to the security challenges of this new model.
The company’s technical team also highlighted threats such as prompt injection, sensitive data exposure, and the complexity of managing multiple agents within a single organization.
Quantum computing and cryptographic transition.
Another focus of the session was the potential impact of quantum computing on current cryptographic systems. Palo Alto Networks warned that the industry could face a transition toward post-quantum cryptography within an estimated horizon of three to ten years, although some scenarios could accelerate that timeline.
The company noted that it is already developing capabilities to identify cryptographic assets across organizations and facilitate migration toward standards more resilient to future attacks.
APIs, SaaS, and the future of software.
Arora also addressed the impact of AI on the enterprise software ecosystem, including the role of APIs and the future of SaaS models. In his view, the adoption of agents could change how enterprise systems interact, reducing reliance on traditional interfaces and accelerating process automation.
He suggested that this shift could progressively transform the current software-as-a-service model, as more tasks are executed by agents rather than human users through graphical interfaces.
Event context.
Palo Alto Networks indicated that these developments are part of its broader strategy to anticipate the evolution of artificial intelligence in enterprise environments, focusing on securing autonomous systems and managing new risk vectors associated with their adoption.
|
|
