At a Glance: Where Your Data Meets AI

This table summarizes every Tenable AI-powered feature described herein.

Feature	AI/ML Technology	Customer data usage	Third-party LLM usage
Vulnerability Priority Rating (VPR)	Machine Learning / Generative AI	No: scores are computed from public and commercial threat intelligence, not your environment	Yes, for the generative-AI component that reads news and advisory content (not your data)
CVSS Metrics Prediction	Machine Learning	No	No
Operating System Prediction	Machine Learning	Yes: your scan data (network-level host metadata only)	No
User Criticality Rating	Machine Learning	Yes: your users’ job titles and access rights	No
AI Threat and Misuse Detection	Deep Learning / Small Language Models	Yes: activity on your connected AI platforms	No (Tenable-trained models)
Data Exposure Detection	Deep Learning / Small Language Models	Yes: content passing through your connected AI platforms	No (Tenable-trained models)
Prompt Translation	Deep Learning / Small Language Models	Yes: prompts you submit in a non-English language	No (Tenable-trained models)
Attack Path Summary, Assistant, and Mitigation Guidance	Generative AI	Yes: your attack path, asset, and vulnerability data	Yes (Google Vertex AI)
Asset Summary and Natural Language Search	Generative AI	Yes: your asset inventory	Yes (Google Vertex AI)
Exposure Signals Description and Summary	Generative AI	Yes: your exposure signal data	Yes (Google Vertex AI)
Tenable Hexa AI	Generative / Agentic AI	Yes: your asset, vulnerability, identity, configuration, and AI-system data already in Tenable, plus what you type	Yes (Anthropic, via AWS Bedrock)
Adversary View (Tenable Hexa AI)	Generative / Agentic AI	Yes: your exposure, asset, and vulnerability data already collected in Tenable One	Yes (Anthropic Claude Mythos, via AWS Bedrock)
Tenable Research	Generative / Agentic AI	Yes: your scan data (network-level host metadata only)	Yes, (Anthropic, via Claude Code)