Intel worked with LayerX to explore on-device, AI-powered security use cases on IntelĀ®ļø Coreā¢ļø Ultra AI PCs, leveraging the integrated NPU (Neural Processing Unit) that is powering many enterprise endpoints worldwide.
Tested and Trusted
Exploring On-Device AI Security on Intel-powered AI PCs. As part of ongoing work to explore next-generation AI PC experiences, Intel worked with LayerX to evaluate on-device, AI-powered security solutions on IntelĀ® Coreā¢ Ultra processors. The team examined how LayerXās browser-based detection models operate locally using available platform capabilities, publishing the results.Ā
For enterprise security leaders, the takeaway is simple: this is productionāgrade AI usage security, already deployed in the browser today.
Intelās engineering team evaluated and benchmarked LayerX's local AI workloads on IntelĀ® Coreā¢ Ultra X7 358H against competing silicon.Ā
LayerX leverages IntelĀ® Coreā¢ Ultra NPU to run specializedĀ language models directly in the browser ā real-time AI usage detection with no cloud round-trip.
In Intel's testing, LayerX ran up to **2x faster** on IntelĀ® Coreā¢ Ultra X7 358H than on competing processors, across data classification, summarization, and phishing detection workloads.
Generative AI has rewritten the rules of enterprise security. See the full story behind the engineering work, the benchmark methodology, and what on-device AI security means for the AI PC era.
Cloud-based AI enforcement breaks down where it matters most. Routing every user prompt to a remote model means sending sensitive data off the device, accepting latency that defeats real-time enforcement, and depending on network uptime exactly when you can't afford to.
The next generation of security has to live where the user lives ā in the browser, at the moment of interaction, on the endpoint itself. Intel worked with LayerX to prove production-grade AI security can run there, without compromise.
LayerX's on-device small language model understands meaning, not just keywords ā recognizing sensitive business IP that legacy DLP can never catch. All inference happens locally on the Intel NPU.
Prompt injection, jailbreaking, and guardrail manipulation demand AI-native defenses. LayerX identifies these patterns in real time, in the browser, as they unfold.
No prompts leave the device. No content traverses the network for inspection. Powered by IntelĀ® Coreā¢ Ultra, LayerX delivers AI security that is private by design ā not by policy.
"Intel is working with LayerX to advance AI PC security capabilities that bring new levels of visibility and enforcement for the modern, AI-driven workforce. With Intel's WebGPU frameworks, IntelĀ® Coreā¢ Ultra 3 delivers up to 2x faster response times versus AMD Ryzen AI ā an advantage that becomes critical as enterprise browsers increasingly inspect all user and agent interactions."
LayerX runs on the IntelĀ® Coreā¢ Ultra-powered AI PCs already standardized across your fleet. No new hardware, no procurement cycle.
On-device inference means every user, every prompt, every browser session is inspected at the moment of action.
A lightweight browser extension. No agents, no infrastructure rollout. Deploy across thousands of endpoints in days.
