See What Our AI Content Auditor Finds
Real-world sample outputs from our auditing tools. Every report, score, and finding is generated automatically.
📋 Audit Report Sample REPORT
Q4 Marketing Strategy — Acme Corp
72
Overall Accuracy Score72 / 100 — Needs Review
Critical Findings
CRITICAL
Fabricated statistic: “87% of consumers prefer AI-generated content”
No source provided. No matching study found in academic or industry databases.
CRITICAL
Hallucinated citation: “Harvard Business Review, March 2025”
This issue of HBR does not exist. Publication verified against HBR archive.
CRITICAL
Inflated ROI claim: “$4.2M projected savings”
No methodology, model, or supporting calculation provided for this projection.
High Findings
HIGH
Outdated market data — 2021 figures cited as current
Section 3.2 references market share data from 2021 without noting age of data.
HIGH
Misattributed quote
Quote attributed to CEO of TechCorp was actually made by their VP of Marketing.
HIGH
Inconsistent metrics between sections
Executive summary states 32% growth; Section 4.1 reports 28% for the same period.
HIGH
Unverified competitor comparison
Claims “3x faster than CompetitorX” with no benchmark or testing methodology cited.
14
Claims Checked
7
Flagged
3
Critical
4
High
72%
Accuracy
🛡 Compliance Report Sample HIPAA
HIPAA Compliance Audit — MedFirst Health Systems
61
Compliance Score61 / 100 — Needs Review
Critical Flags
FAIL
PHI exposure in AI-generated patient summary
Patient full name and date of birth appear in an AI-generated discharge summary without de-identification.
Remediation: Implement PHI detection and automatic redaction in the AI generation pipeline before output is stored or displayed.
FAIL
Unencrypted data reference in recommendation
AI recommendation output references lab results via a plain-text URL without TLS enforcement.
Remediation: Enforce HTTPS for all internal data references. Update AI output templates to use encrypted endpoints only.
High Flags
FLAG
Missing Business Associate Agreement (BAA) reference
The AI vendor integration section does not reference an executed BAA.
Remediation: Confirm BAA is in place with AI provider and reference it in system documentation.
FLAG
Inadequate access control language
Policy states “authorized personnel may access records” without defining role-based access controls.
Remediation: Define specific RBAC roles and map them to minimum necessary access levels.
FLAG
Retention period not specified
No data retention or disposal policy is referenced for AI-generated clinical summaries.
Remediation: Establish and document a retention schedule consistent with HIPAA §164.530(j) (6-year minimum).
| Check | Status | Details |
|---|---|---|
| PHI De-identification | FAIL | PHI found in AI output |
| Encryption at Rest | PASS | AES-256 confirmed |
| Encryption in Transit | FAIL | Plain-text URL detected |
| BAA Documentation | FLAG | Not referenced |
| Access Controls | FLAG | Insufficient specificity |
| Audit Logging | PASS | Comprehensive logs enabled |
| Retention Policy | FLAG | Not documented |
| Breach Notification | PASS | Process documented |
| Staff Training | N/A | Outside audit scope |
🏠 Underwriting Validation Report LENDING
Residential Appraisal — 1847 Oak Valley Dr
84
Validation Score84 / 100 — Good
Property Valuation
| Metric | Value | Status |
|---|---|---|
| Appraised Value | $485,000 | |
| AVM Estimate | $462,000 | |
| Variance | 4.9% | FLAG |
FLAG
Appraised value exceeds AVM estimate by 4.9%
Variance above 4% threshold triggers manual review. Recommend additional comparable analysis.
Income & Asset Verification
| Document | Amount | Cross-Check | Status |
|---|---|---|---|
| W-2 Income | $87,400 | Matches tax return | ✓ |
| Tax Return (AGI) | $87,400 | Matches W-2 | ✓ |
| Bank Statements (3-mo avg) | $24,892 | Sufficient for down payment | ✓ |
Comparable Sales Analysis
| Comp | Sale Price | Distance | Status |
|---|---|---|---|
| Comp #1 | $478,000 | 0.4 mi | PASS |
| Comp #2 | $491,500 | 0.7 mi | PASS |
| Comp #3 | $469,000 | 2.1 mi | FAIL |
RED FLAG
Comparable sale #3 is 2.1 miles from subject property
Exceeds the 1-mile standard for residential comparables. This comp may not be representative of the local market.
Regulatory Compliance
| Regulation | Status | Notes |
|---|---|---|
| RESPA | PASS | All disclosures present and timely |
| TILA | PASS | APR and finance charges correctly disclosed |
| QM / ATR | FLAG | DTI at 44.8% — exceeds 43% threshold |
🔄 Before / After Accuracy Comparison DIFF
Before Audit
“Our AI chatbot reduced support tickets by 73%.”
FAIL — No source. Fabricated metric with no internal data reference.
After Audit
“Our AI chatbot reduced support tickets by 38% (Source: Internal Q3 2025 analysis, n=4,200 tickets).”
PASS — Verified. Properly cited with sample size.
Before Audit
“According to a 2024 MIT study, large language models are 99% accurate.”
FAIL — Hallucinated citation. No such MIT study exists.
After Audit
“According to Liang et al. (2023), “Holistic Evaluation of Language Models,” Stanford CRFM, LLM accuracy varies by task (42–91%).”
PASS — Real citation. Nuanced claim with range.
Before Audit
“We are the industry-leading platform with the highest customer satisfaction.”
FAIL — Superlative claims with no benchmark or survey data.
After Audit
“Our platform achieved a 92 NPS score in Q2 2025 (n=1,840 respondents), ranking in the top quartile per G2 Grid Report.”
PASS — Specific metric, sample size, and third-party benchmark.
Before Audit
“Our solution saves companies $2M+ annually.”
FAIL — Unsubstantiated financial claim. No case study or methodology.
After Audit
“In a pilot with 3 enterprise clients, our solution reduced manual review costs by $340K–$580K over 12 months (see Case Study CS-2025-07).”
PASS — Specific range, defined scope, and linked case study.
🚩 Red Flag Examples FLAGS
CRITICAL Fabricated Citation
“According to a 2024 Stanford study on enterprise AI adoption...”
No matching study found in Stanford publications, SSRN, or Google Scholar. Citation does not exist.
CRITICAL Hallucinated Statistic
“Our model achieves a 95% accuracy rate across all benchmarks.”
No supporting data, test results, or benchmark references provided. Claim is unverifiable.
HIGH Outdated Data
“The market grew 12% in 2023, driven by cloud adoption.”
Source data is from a 2021 report. The 12% figure refers to 2020–2021 growth, not 2023.
HIGH Misattributed Quote
“As Satya Nadella said, ‘AI will replace 40% of jobs by 2030.’”
This quote was never made by Satya Nadella. It is a paraphrase of an unrelated McKinsey estimate, misattributed.
MEDIUM Inconsistent Numbers
Executive summary: “$2.1M in savings.” — Section 4: “$1.8M in savings.”
The same metric is reported with two different values in the same document. Indicates copy-paste error or hallucination.
MEDIUM Unverified Claim
“Industry-leading performance with best-in-class uptime.”
No benchmark, SLA reference, or third-party validation. Superlative marketing language without evidence.
📈 Accuracy Score Examples SCORES
EXCELLENT (90–100)
API Documentation v3.2
94
GOOD (80–89)
Annual Financial Report
86
NEEDS REVIEW (60–79)
Marketing Whitepaper
68
CRITICAL (0–59)
AI-Generated Blog Post
41
Score Distribution
8%
0-59
27%
60-79
41%
80-89
24%
90-100
Distribution across 2,400+ documents audited
Scoring Methodology
- Claims Verified — Every factual claim is cross-referenced against source material, public databases, and academic repositories.
- Sources Checked — Citations are validated for existence, recency, and relevance. Broken links and non-existent publications are flagged.
- Consistency Analysis — Numbers, dates, and metrics are compared across all sections of the document for internal agreement.
- Freshness of Data — Data points are checked against their original publication date. Outdated figures presented as current are penalized.
// Continue Exploring