AI Security · On-device

Starting mobile or Angular development with on-device models?

Pick your platform and model, and get a practical starter guide: runtime options, four ways to deliver the model safely, and a security and coding checklist you can work through with your team. Runs in your browser; nothing you select is uploaded.

Organised around the NIST AI RMF, the 2025 OWASP GenAI/LLM Top 10, MITRE ATLAS and the MAESTRO multi-agent framework — referenced by name, with our own wording throughout.

Building this with the LyboAI Edge Runtime?

The Edge Runtime is our on-device agent engine for exactly this pattern. Signed packs, least-privilege tools and local execution cover a good share of this checklist out of the box.

Explore the Edge Runtime Assess the use case first

Framework references

NIST AI Risk Management Framework Govern, Map, Measure, Manage — the enterprise governance backbone.OWASP GenAI / LLM Top 10 (2025) The application-security categories referenced as LLM01–LLM10.MITRE ATLAS The adversarial threat landscape for ML pipelines.MAESTRO (Cloud Security Alliance) Threat modelling for agentic, multi-agent systems.

These frameworks are referenced by name as sources of the categories used; every checklist item is LyboAI's own wording. Indicative guidance for teams getting started, not a certification, an audit, or a substitute for your security review.