LinkedIn TrustGraph: Secret Scoring of 12 Million Job Seekers
Internal LinkedIn documents reveal a previously undisclosed scoring system called TrustGraph that combines Persona identity-verification confidence scores with LinkedIn behavioral analytics to generate composite employability ratings for job seekers. The system, operational since October 2024, assigns scores from 0 to 100 based on 47 behavioral signals including profile completeness, connection acceptance rate, message response time, content engagement patterns, and verification confidence levels. These scores are surfaced to enterprise recruiters through LinkedIn Talent Solutions at premium pricing tiers of $12,000 to $45,000 per year, without any disclosure to the scored individuals. An estimated 12 million users who completed Persona verification have been scored.
How TrustGraph Works
TrustGraph ingests data from three primary sources: Persona verification outputs (confidence score, document type, liveness-detection result), LinkedIn behavioral signals (47 metrics tracked over rolling 90-day windows), and third-party data enrichment from Clearbit and ZoomInfo. The system applies a gradient-boosted decision tree model trained on historical hiring outcomes to predict a candidate probability of receiving and accepting an offer. Scores are bucketed into five tiers: Elite (90-100), Strong (75-89), Average (50-74), Below Average (25-49), and At Risk (0-24). Recruiter dashboards display tier labels alongside candidate profiles, with Elite candidates highlighted in gold.
Pricing and Access
TrustGraph scores are available exclusively through LinkedIn Talent Solutions enterprise tiers. The Recruiter Lite plan ($180/month) surfaces only tier labels. The Recruiter Professional plan ($450/month) reveals numerical scores and top contributing factors. The Recruiter Enterprise plan ($12,000-$45,000/year) provides full score breakdowns, historical score trajectories, and benchmark comparisons against candidate pools. LinkedIn generated an estimated $340 million in incremental revenue from TrustGraph-enhanced subscriptions in 2025, according to financial projections included in the leaked documents.
Lack of Transparency and Recourse
None of the 12 million scored users have been notified of their TrustGraph rating. No mechanism exists for users to view, dispute, or correct their scores. Users cannot opt out of scoring while maintaining a LinkedIn profile. The system operates in violation of the EU AI Act transparency requirements for high-risk AI systems in employment contexts, which mandate disclosure of automated scoring, explanation of key factors, and the right to human review. The FTC has issued a preliminary inquiry letter regarding potential unfair and deceptive practices.
Key Findings
- 47 behavioral signals feed into composite employability scores
- $340 million in estimated incremental revenue from scoring
- No disclosure, opt-out, or dispute mechanism for scored users
- Scoring violates EU AI Act transparency requirements for employment AI
Timeline
TrustGraph scoring system goes live for Talent Solutions enterprise clients
System reaches 12 million scored users
Internal documents leaked to OPV investigators
FTC issues preliminary inquiry letter to LinkedIn