Skip to content
Cases
Investigation

Company Alpha

Contrast requiredHigh riskExternal observation·Owner M. OrtegaDemo · Sample investigation
Open report
  1. ObserveCompleted
  2. Detect8 claims
  3. 3ContrastIn progress
  4. 4VerifyPending
Evidence collected
142
97.9% intact
Claims detected
8
3 verified
Contradictions
2
Need explanation
Unresolved
3
Awaiting contrast
Evidence map
CLM-001 · every source behind the claim
All claims
CLM-001 · Claim

All automated credit decisions are reviewed by a human before final approval.

Source: Public AI policy, §4.2

Contradicted
Evidence strength84%
  4. OrganizationOrganization responseAwaiting response
  5. InternalInternal decision logsRequires authorization
Verification state
  • Supporting1
  • Contradicting2
  • OrganizationAwaiting response
  • InternalRequires authorization
  • Supports
  • Contradicts
  • Organization response
  • Needs authorization
Overall status
Contrast requiredSome findings need an answer from the organization.
Claims by outcome8 total
  • Verified3
  • Partially verified1
  • Contradicted2
  • Insufficient evidence1
  • Unresolved1
Evidence integrity
97.9%Intact
139 of 142 hashes verified
12 sampled in view
Evidence coverage
68%

Share of claims backed by at least one intact piece of evidence.

View evidence

Claims

All claims

Contradictions2

  • CLM-001
    All automated credit decisions are reviewed by a human before final approval.
    2 evidence items contradict this claim.
  • CLM-004
    Training data excludes customer-identifiable information.
    1 evidence item contradicts this claim.

Open questions5

  1. 01Which step, if any, reviews credit decisions before approval?
  2. 02How many automated approvals in the last quarter had a reviewer record?
  3. 03Which datasets were used to train the assistant, and how were they filtered?
  4. 04Where are model outputs stored, and who can read them?
  5. 05Which third-party models are in production and when were they assessed?

Human review required4

  • CLM-001Escalated
    All automated credit decisions are reviewed by a human before final approval.
    M. Ortega
  • CLM-003In review
    Model outputs are logged and retained for 24 months.
    A. Okafor
  • CLM-004In review
    Training data excludes customer-identifiable information.
    M. Ortega
  • CLM-005In review
    Fraud-detection model is re-evaluated against a fixed benchmark every quarter.
    S. Marlowe

Timeline

  1. Contrast request prepared
    3 questions on credit decision review ready to send.
    Today, 10:48
  2. CLM-001 marked contradicted
    Timing and outcome evidence conflict with the policy claim.
    Today, 09:12
  3. Reviewer escalation
    M. Ortega escalated CLM-001 for senior review.
    Yesterday, 17:30
  4. Organization document received
    EV-0006 uploaded through secure channel.
    Oct 4, 11:08
  5. Credit decision timing captured
    240 controlled submissions analyzed.
    Oct 3, 14:02
  6. Observation started
    8 claims detected from public sources.
    Oct 1, 15:30
Dashboard — V1LA