Public Figure Fact Checker
Identify potential misinformation and fact-check claims made by public figures with structured true/false output and correction recommendations.
When to use
Trigger when users request:
- "Fact-check this document"
- "Verify these specifications"
- "Check if this information is still accurate"
- "Update outdated data in this claim"
- "Validate the claims in this section"
Workflow
Copy this checklist to track progress:
Public Figure Fact-Check Progress:
- [ ] Step 1: Extract verifiable claims
- [ ] Step 2: Search Google Fact Check Tools (MANDATORY)
- [ ] Step 3: Search additional authoritative sources
- [ ] Step 4: Generate structured fact-check report
Step 1: Extract verifiable claims
Identify specific factual statements made by the public figure:
Target claim types:
- Statistical data and numbers
- Historical events and dates
- Scientific or medical statements
- Financial figures and company data
- Policy positions and voting records
- Biographical information
Extract format:
CLAIM: [Exact quote from public figure]
CONTEXT: [When/where it was said]
TYPE: [Statistical/Historical/Scientific/etc.]
Step 2: Search Google Fact Check Tools (MANDATORY)
REQUIRED: Always start with Google Fact Check Tools before other sources
- Google Fact Check Explorer: Search for the exact claim or similar statements
- Review existing fact-checks: Check if the claim has been previously verified
- Analyze ClaimReview data: Look for structured fact-check results
- Cross-reference sources: Note which fact-checkers have covered this topic
If Google Fact Check Tools shows existing fact-checks:
- Review the verdict and methodology
- Check if the context matches current claim
- Verify the credibility of the fact-checking organization
- Use as primary reference point for your analysis
If no existing fact-checks found:
- Proceed to Step 3 for original verification
- Note in report that this is a new claim requiring fresh fact-checking
Step 3: Search additional authoritative sources
For each claim, search official and credible sources:
Government data:
- Official statistics (census.gov, bls.gov, cdc.gov)
- Government databases and reports
- Congressional voting records
- Court documents and legal filings
Academic and research:
- Peer-reviewed studies
- University research institutions
- Scientific journals and publications
- Medical organizations (WHO, CDC, FDA)
Financial and business:
- SEC filings and annual reports
- Financial news from established outlets
- Company press releases
- Stock market data
Historical verification:
- Historical archives and records
- Newspaper archives
- Documentary evidence
- Timeline verification from multiple sources
Step 4: Generate structured fact-check report
Output format for each claim:
## FACT-CHECK RESULT
### CLAIM #[N]
**Statement:** "[Exact quote]"
**Source:** [Public figure name] on [date/platform]
**Category:** [Statistical/Historical/Scientific/Financial/Policy]
**VERDICT:** ✅ TRUE / ❌ FALSE / ⚠️ MISLEADING / ❓ UNVERIFIABLE
**VERIFICATION:**
- **Google Fact Check Result:** [Run: python scripts/google_fact_checker.py "claim text"]
- **Primary Source:** [Authoritative source URL]
- **Supporting Sources:** [Additional verification sources]
- **Date Checked:** [Current date]
**ANALYSIS:**
[Brief explanation of why the claim is true/false/misleading]
**CORRECTION (if applicable):**
**Accurate Statement:** "[Corrected version of the claim]"
**Key Difference:** [What was wrong/misleading about original]
Source Evaluation Guidelines
Authoritative Sources (High Priority)
- Google Fact Check Tools - MANDATORY first step for all claims
- Government agencies - Official statistics, records, databases
- Academic institutions - Peer-reviewed research, university studies
- Medical organizations - WHO, CDC, FDA, medical journals
- Financial regulators - SEC filings, central bank data
- News organizations - Established outlets with fact-checking standards
- Other fact-checking platforms - Snopes, PolitiFact, FactCheck.org (as secondary verification)
Use with Caution
- Corporate press releases (may contain bias)
- Think tank reports (check funding sources)
- Social media posts (even from experts)
- Single-source claims
Red Flags
- No authoritative source available
- Google Fact Check Tools returns contradictory results
- Sources contradict each other significantly
- Only partisan sources support the claim
- Information is very recent with limited verification
CRITICAL RULE: Never complete a fact-check without consulting Google Fact Check Tools first
Utility Scripts
To automate Google Fact Checker API requests, use the simplified script:
Simple Google Fact Check Query
python scripts/google_fact_checker.py "Senator X said unemployment dropped 40% during his term"
Returns raw string format:
- Multiple lines for top 5 results:
#N: InformationContent NO_RESULTS- if no existing fact-checksERROR: message- if API error occurred
Example outputs:
#1: Donald Trump was born in the United States | Birth certificate confirms he was born in New York City | Original claim: Trump was born in USA
#2: Official records verify Trump's birthplace as New York | Original claim: Donald Trump birthplace question
#3: Birth certificate authentic and valid | Multiple sources confirm NYC birth
NO_RESULTS
ERROR: API key not found
Examples
Example 1: Political Statistical Claim
User request: "Fact-check: 'Senator X said unemployment dropped 40% during his term'"
Output:
## FACT-CHECK RESULT
### CLAIM #1
**Statement:** "Unemployment dropped 40% during my term in office"
**Source:** Senator X on Twitter, January 8, 2026
**Category:** Statistical
**VERDICT:** ⚠️ MISLEADING
**VERIFICATION:**
- **Google Fact Check Result:** #1: Unemployment claim misleading - shows 2.1 percentage point decrease, not 40% absolute drop | Original claim: Unemployment dropped 40% during term
#2: Economic statistics misrepresented in political statement | Original claim: Similar unemployment claim
- **Primary Source:** Bureau of Labor Statistics unemployment data
- **Supporting Sources:** Federal Reserve economic data, state employment records
- **Date Checked:** January 11, 2026
**ANALYSIS:**
Unemployment did decrease during the senator's term, but only by 2.1 percentage points (from 5.3% to 3.2%), which represents a 40% relative decrease but not an absolute 40% drop.
**CORRECTION:**
**Accurate Statement:** "Unemployment decreased by 2.1 percentage points (a 40% relative reduction) during my term"
**Key Difference:** Original statement implies much larger absolute decrease than actually occurred
Example 2: CEO Business Claim
User request: "Verify: 'Our company's revenue grew 500% last year'"
Output:
## FACT-CHECK RESULT
### CLAIM #1
**Statement:** "Our company's revenue grew 500% last year"
**Source:** CEO Jane Doe at investor conference, January 2026
**Category:** Financial
**VERDICT:** ✅ TRUE
**VERIFICATION:**
- **Google Fact Check Result:** NO_RESULTS
- **Primary Source:** Company's SEC 10-K filing
- **Supporting Sources:** Annual report, audited financial statements
- **Date Checked:** January 11, 2026
**ANALYSIS:**
SEC filings confirm revenue increased from $2M in 2024 to $12M in 2025, representing exactly a 500% increase.
**CORRECTION:** None needed - claim is accurate.
Quality checklist
Before completing fact-check:
- All factual claims extracted with exact quotes
- Google Fact Check Tools consulted FIRST (MANDATORY)
- Each claim verified against authoritative sources
- Verdict clearly stated (TRUE/FALSE/MISLEADING/UNVERIFIABLE)
- Sources are credible and current
- Analysis explains the reasoning
- Corrections provided for false/misleading claims
- Temporal context included where relevant
Limitations
This skill cannot:
- Verify future predictions or speculation
- Determine subjective truth in opinion-based statements
- Access private or confidential information
- Resolve disputes where authoritative sources disagree
- Complete fact-checks without using Google Fact Check Tools
For such cases:
- Mark as ❓ UNVERIFIABLE
- Note the limitation in analysis
- Suggest seeking additional expert consultation
- Always include Google Fact Check Tools results even if no matches found
