Lately, he stated, there have been information reviews that AI brokers created by companies prompted hacks inside their very own corporations. He didn’t cite particular examples, however last week Meta said there had been a extreme inside safety breach after an autonomous AI agent uncovered delicate firm and person information to unauthorized workers for 2 hours.
Sooner or later, if brokers within the enterprise are greater than a fad, Arora stated, “there will probably be hundreds of thousands of brokers traversing enterprise architectures, making an attempt to execute on their behalf — each brokers delegated by individuals such as you and me, and autonomously. I can’t think about assembly a CEO within the final three months who doesn’t have some aspiration to start out having brokers successfully doing duties inside the enterprise. It’s sluggish going, however the intention is there. And I can see many system integrators and consultants on the market advocating and serving to clients with that migration.”
However, he added, there are dangers. To satisfy them, Prisma AIRS 3.0 will permit admins to securely deploy AI purposes, he stated. To extend visibility, the platform will establish brokers working in cloud environments, on SaaS platforms and domestically on endpoints. A functionality known as Agent Artifact Safety maps out an agent’s structure and scans for vulnerabilities, and one other functionality known as AI Purple Teaming for Brokers simulates context-aware agentic assaults, discovers AI-related vulnerabilities, and recommends runtime safety insurance policies.
Prisma Browser
To additionally enhance AI safety, Palo Alto Networks launched a brand new model of Prisma Browser for enterprise finish customers, with expanded capabilities permitting workers to make use of any LLM they select. The brand new model of the browser is ready to uncover user-generated AI exercise and implement content-aware boundaries to maintain brokers inside their supposed scope. The browser additionally prevents delicate information from leaking to unmanaged or public AI instruments throughout automated duties, identifies and blocks immediate injection assaults, together with malicious directions designed to hijack AI brokers hidden inside web sites.
Palo Alto Networks stated the browser additionally offers real-time distinction between human actions and automatic AI duties. By assessing the intentions of each human and non-human identities, Prisma Browser allows whole accountability and compliance with evolving world AI rules
Subsequent Technology Belief Safety
Individually, Palo Alto Networks additionally introduced a brand new digital certificates lifecycle administration platform, following the closing final month of its acquisition of CyberArk.
