PII detection

Detect sensitive data before it spreads.

PII detection answers the next question after intent: What sensitive data is present? It helps teams see when AI interactions involve personal information, secrets, or restricted material.

SD-1 / PII detection Detect Classify Evaluate Customer ID Email Token Restricted record Warn In policy Review Policy issue Evidence retained for the policy decision
SD-1: Capability-specific hero showing sensitive entities highlighted before external AI submission. Production metadata: 3ls-sd-1-solutions-pii-detection-composite.svg; SVG master rendered fluidly with explicit 1600x1200 dimensions; crop-safe for 4:3 and stacked mobile hero placement.; task-1093-inline-svg-master

What sensitive data is present?

3LS gives operators a readable view of when prompts, tool inputs, and outputs include customer data, credentials, or other content that needs extra care.

Phase 1

Sensitive content appears

A prompt, upload, tool input, or output includes personal data, credentials, or restricted material.

Phase 2

Data types are identified

3LS highlights what kind of sensitive content is present so teams can separate ordinary work from data-risk events.

Phase 3

Policy context is applied

The interaction can be warned, blocked, reviewed, or allowed based on sensitivity and business context.

Phase 4

Evidence is retained

Operators keep a readable record of what was detected and what decision followed.

SD-2 / Detect, classify, evaluate Detect Classify Evaluate Activity Analysis Policy result Evidence In policy Review Policy issue Evidence retained for the policy decision
SD-2: Three-step visual showing observed activity, analysis, then policy evaluation and evidence. Production metadata: 3ls-sd-2-solutions-policy-flow-diagram.svg; SVG master rendered fluidly with explicit 1600x900 dimensions; usable as 16:9 desktop panel and full-width mobile diagram.; task-1093-inline-svg-master

Personal data

Highlight customer and employee information

Surface names, identifiers, and other personal information when people paste, upload, or transform sensitive material with AI.

Outcome: Reduced accidental exposure

Secrets and credentials

Catch the details that should not move further

Spot tokens, credentials, and similar material inside prompts, tool inputs, and outputs before they spread into the wrong place.

Outcome: Fewer high-impact mistakes

Restricted content

Recognize when business-sensitive material is involved

Bring attention to restricted records and internal material so teams can decide whether to continue, warn, or stop the interaction.

Outcome: Clearer data protection decisions

Operator view

Turn AI interactions into clear decisions

Give security teams a readable trail of what was detected, how policy was evaluated, and which evidence supported the result.

Operator evidence view highlights PII in a customer CSV upload and records a policy evaluation with its rationale.

Recent findings

Intent, sensitivity, and policy evaluation

3 evaluations captured
Activity
Intent
Sensitivity
Policy result
Customer records pasted into an assistant
Data handling
PII detected
Needs review
Prompt requests external tool access
Tool use
No sensitive data
Within policy
Prompt includes an API token and a request to share it
Data sharing
Secret detected
Escalate

From the blog

PII Exposure and Detection

Coverage on sensitive-data exposure in AI workflows, where copied records leak, and why runtime detection matters before the prompt leaves the browser.