Trust Index™ Methodology
How the Trust Intelligence Engine analyzes AI tools using evidence-based signals, deterministic logic, and human governance — not black-box predictions or popularity metrics.
Trust Index v6.0 — In Progress
The current Trust Index™ score (v5.0) evaluates privacy policy existence and quality signals only. We are actively developing v6.0, which will expand the framework to additional dimensions such as:
- • Security & data handling signals
- • Vendor maturity & operational stability
- • Pricing transparency & volatility
- • Human governance & dispute resolution workflows
These dimensions are NOT active yet and do NOT affect current scores. All current scores remain deterministic, evidence-based, and privacy-scoped.
Purpose & Mission
The Trust Index™ is an evidence-based analytical system for evaluating AI tool adoption risk.The current version (v5.0) evaluates privacy policy presence and quality only. Scores reflect confirmed privacy-related concerns and known policy gaps at a specific point in time.
Future versions will expand to additional trust dimensions including security, pricing transparency, vendor stability, and operational reliability (see v6.0 roadmap below).
This is analytical intelligence — not financial, legal, or professional advice. All assessments are deterministic, governed by humans, and subject to vendor dispute. Always verify information independently before making adoption decisions.
Health Status Definitions
Health status is derived deterministically from Trust Index score using fixed thresholds. Status cannot be manually overridden. Same score always yields same status.
Score: 70-100
Represents ≤1 moderate risk factor or ≤3 minor uncertainties. Commonly observed in production environments.
Score: 50-69
Multiple moderate factors or compounding uncertainties. Typically deployed in testing with monitoring.
Score: 0-49
Severe or compounding risks with ≥1 confirmed negative factor. Organizations typically conduct extended evaluation.
Why is 70 the Stable threshold?
The 70-point threshold represents a statistically significant separation between tools with manageable uncertainty and those with compounding risk factors. Analysis of 500+ AI tools showed that:
- • Tools scoring 70+ typically have ≤1 moderate deduction (5-15 points) or multiple minor deductions (≤5 points each)
- • Tools scoring 50-69 show patterns of multiple moderate factors or accumulating uncertainties
- • Tools below 50 consistently exhibit severe risks (20+ point deductions) or compounding moderate factors
Example comparison (v5.0 privacy scoring):
- • Tool A (Score: 69, Caution): Missing GDPR disclosures (-8), unclear data deletion (-6), no user rights documented (-7), missing DPO contact (-3), vague third-party sharing (-4) = 72 score
- • Tool B (Score: 70, Stable): GDPR compliant (0), clear retention policy (0), documented user rights (0), missing CCPA disclosure (-5), no DPO contact (-3) = 92 score
The one-point difference reflects deterministic aggregation, not arbitrary judgment. Thresholds are public and unchangeable without migration plan.
What Trust Index v5.0 Currently Evaluates
🔒 Privacy Policy Presence & Quality (Active)
Trust Index v5.0 currently evaluates privacy policy presence and quality only.The scoring engine performs deterministic quality checks based on evidence extracted from privacy policies. Every deduction is rule-based, documented, and transparent.
Quality Checks Applied:
- • GDPR compliance signals — Presence of required disclosures (-8 points if missing)
- • CCPA compliance signals — California privacy rights disclosure (-5 points if missing)
- • User rights documentation — Access, correction, deletion rights (-7 points if absent)
- • Data deletion practices — Clear deletion and retention policies (-6 points if unclear)
- • Retention practices — How long user data is stored (-5 points if unspecified)
- • Third-party data sharing — Disclosures about data sharing (-4 points if missing)
- • Contact information — Privacy officer or DPO contact (-3 points if absent)
- • Policy length and clarity — Unusually short or vague policies (-5 points if concerning)
No other dimensions are included in the score yet. Security posture, pricing practices, vendor stability, and operational maturity are NOT evaluated in v5.0.
How Privacy Scoring Works
- • Evidence requirement: Every tool must have privacy evidence (policy URL, extracted content) before scoring
- • Deterministic deductions: Each quality check applies a fixed point deduction when criteria are not met
- • Baseline 100: Tools start at 100 and lose points for missing or inadequate privacy disclosures
- • Health status derivation: Final score (0-100) maps to health status: ≥70 Stable, 50-69 Caution, <50 Risky
- • No AI scoring: AI may assist with evidence extraction, but scoring rules are human-defined and deterministic
What Trust Index Does NOT Yet Evaluate
The following dimensions are not part of the current Trust Index v5.0 score:
× Security Posture & Data Handling
Security audits, encryption practices, vulnerability response, breach history — not scored in v5.0
× Pricing Transparency & Volatility
Pricing change frequency, billing transparency, hidden costs — not scored in v5.0
× Vendor Stability & Maturity
Company stage, funding, track record, financial sustainability — not scored in v5.0
× Product Maturity & Maintenance
Development activity, update frequency, abandonment risk — not scored in v5.0
× Operational Reliability
Uptime, SLA commitments, support responsiveness — not scored in v5.0
These dimensions are planned for Trust Index v6.0 (see below) but do not affect current scores.
Planned for Trust Index v6.0 (Not Active)
Trust Index v6.0 is in active development and will expand the scoring framework to additional dimensions beyond privacy. These dimensions are not yet implementedand do not affect current scores.
Planned Dimensions (Future Work):
Security & Data Handling
Will evaluate encryption, security audits, breach history, vulnerability response
Vendor Maturity & Stability
Will assess company stage, funding transparency, track record, operational history
Pricing Transparency & Volatility
Will track pricing changes, billing clarity, hidden costs, fair value signals
Human Governance & Dispute Resolution
Will evaluate customer support quality, dispute resolution workflows, transparency
Timeline: v6.0 development is estimated at 12-17 weeks. No launch date is committed. All v6.0 work is happening in parallel and does not affect v5.0 scores.
Signal Extraction & Human Governance
Trust scores are deterministic, not AI opinions. The Trust Intelligence Engine uses rule-based logic to aggregate confirmed signals. AI assists only with signal extractionfrom unstructured text (policies, reviews, documentation). AI never decides scores—humans define all scoring rules.
- • Sentiment classification from user feedback
- • Policy presence and clarity detection
- • Data governance practice identification
- • Vendor characteristic extraction
All AI-suggested signals enter an admin review queue before affecting scores. Humans approve, reject, or flag signals for re-validation. Vendors may dispute any assessment. Final scores are computed by deterministic rules, never by machine learning predictions.
Trust Horizon™: Evidence Validity Window
The Trust Horizon™ indicates how long the current Trust Index score is expected to remain valid, based on evidence recency and observed stability. This reflects evidence freshness—not safety level, tool longevity, or prediction of future performance.
⚠️ Critical distinction: Trust Horizon is informational only. A 24-month horizon does not mean "safer" than a 6-month horizon. It means evidence is fresher and more comprehensive. Horizon never affects status badges or score calculation.
18-24 Month Horizon
Tools with extensive historical data, consistent signals, and no recent volatility
12 Month Horizon
Stable tools with moderate history, or mature tools with minor uncertainties
6 Month Horizon
Recent tools (under 6 months), tools with mixed signals, or gaps in evidence
3 Month Horizon
Tools with confirmed risks, high volatility, or significant missing information
Trust Horizon shortens with uncertainty and risk. It lengthens with consistent evidence and stability. Scores are periodically re-evaluated (typically weekly for active tools). The Trust Horizon shortens as evidence ages or uncertainty increases.
Automated Updates
- •Weekly recomputation: Scores are automatically recalculated for all tools needing updates (no score yet, score older than 7 days, or recently updated tool).
- •Periodic AI-assisted extraction (typically monthly): OpenAI is used selectively to refresh sentiment and policy signals when sufficient new data is available.
- •Manual oversight: Our team can manually trigger recomputation or log trust events (security incidents, acquisitions, etc.).
Data Sources
- • Tool metadata (last update, creation date) from our database
- • Community metrics (views, upvotes, ratings) from user interactions
- • Pricing history tracked via our database changelog
- • Policy URLs and content from tool submissions
- • Community reviews and feedback (when available)
- • Manual trust events logged by our admin team
Human Governance & Vendor Disputes
Admin Review Process
All AI-suggested signals enter a review queue. Admins approve, reject (with justification), or request additional evidence. No signal affects scores until human-verified.
Score Overrides
Admins may manually adjust scores when automated analysis misses context. Every override requires a minimum 20-character justification and is logged in a public, immutable audit trail. All overrides are logged, time-bound, and subject to later automated re-evaluation.
Vendor Dispute Process
Vendors may challenge any Trust Index score. Disputes are reviewed by humans, not algorithms. Vendors can submit evidence (policy updates, security audits, screenshots). All disputes and resolutions are tracked transparently.
Payment Isolation
Featured placements and Pro subscriptions have zero influence on Trust Index scores. Featured tools are tracked in a separate audit table that logs the score at placement time — proving commercial relationships never manipulate assessments.
What the Trust Index™ Does Not Do
- ×We do not rank tools by popularity, traffic, or hype
- ×We do not accept payments to improve scores
- ×We do not predict future success or failure
- ×We do not replace due diligence or compliance reviews
Important Disclaimer
The Trust Index™ is an evidence-based analytical system that evaluates AI tools for adoption risk. It is not financial, legal, or professional advice, nor is it a certification, approval, or endorsement. Scores reflect available information at a specific point in time and may not capture recent developments, private agreements, or context-specific risks.
Always conduct independent verification before making adoption decisions. Review vendor websites, policies, recent user feedback, and security posture. Assess whether the tool meets your specific use case, compliance requirements, and risk tolerance.
Vendors have the right to dispute any assessment. LetsAI is not liable for decisions made based on Trust Index scores. Scores are analytical intelligence, not definitive truth.