Mastering automated fact-checking: APIs and ClaimReview for investigative journalists
Learn how to automate fact-checking using Google APIs, ClaimReview, and MediaReview schema to accelerate digital investigations and OSINT workflows
1. Introduction and context
The speed of disinformation campaigns often outpaces traditional reporting cycles. For investigative journalists, the ability to rapidly identify whether a viral claim has already been debunked or to ensure their own investigations are discoverable by global fact-checking databases is becoming increasingly important. Automated fact-checking tools act as a “force multiplier,” allowing newsrooms to scan thousands of claims against verified databases in seconds.
🔍 1.1. The investigative need
In the early stages of breaking news, recycled misinformation is a primary trap. Journalists need automated systems to:
Prevent redundant work: Avoid spending time debunking a video or image that has already been proven to be from a different year, location, or generated by AI by other news or fact-checking organisations, including members of the Eurovision News Spotlight Fact-Checking and OSINT Network.
Monitor narrative trends: Track how a specific false claim evolves across different regions and languages.
Enhance digital authority: By using structured data like ClaimReview, newsrooms can ensure their verified findings are prioritized by search engines, and now by AI assistants.
1.2. Learning outcomes
Query the Google Fact Check Explorer API to automate claim verification.
Implement ClaimReview and MediaReview schema to make investigations machine-readable.
Integrate third-party fact-checking APIs and collaborative networks into OSINT workflows.
Establish a secure chain of custody for digital debunks.
1.3. Case study hook
Imagine a viral photo circulating on social media appearing to show a high-ranking official in an incriminating or highly unusual situation. Before launching a deep-dive forensic analysis, an automated query across global fact-checking databases and reverse-image search reveals that the exact image was identified as “AI-Generated/Deepfake” by a news agency in a different region weeks ago.
💡 2. Foundational theory and ethical-legal framework
2.1. Key terminology
ClaimReview: A standardized schema (tags) used to identify a fact-check article for search engines.
MediaReview: An emerging schema specifically for tagging manipulated or misrepresented images and videos.
API (Application Programming Interface): A bridge that allows your software to “talk” to a database of fact-checks directly.
⚠️ 2.2. Ethical and legal boundaries
2.2.1. Consent & privacy
The “stop at the login” rule: Automated tools must only access public-facing APIs. Never use automated scripts to bypass paywalls or private group settings (e.g., closed Telegram channels) without explicit legal and ethical clearance.
Sensitivity: Fact-checking private individuals requires a higher threshold of public interest than checking public figures.
2.2.2. Legal considerations
🛑 Disclaimer: This tutorial does not constitute legal advice. Investigative techniques involving automated data retrieval may intersect with Computer Fraud and Abuse Acts (CFAA) or GDPR. Always consult your organization’s legal department before deploying large-scale automated scrapers or API integrations.
🛠️ 3. Applied methodology: step-by-step practical implementation
3.1. Required tools & setup
Google Cloud Console Account: To generate an API key for Fact Check tools.
Fact Check Markup Tool: To generate ClaimReview code without manual programming.
Collaborative Access: Membership in networks like Eurovision News Spotlight for verified information.
👷♀️ 3.2. Practical execution (The “How”)
Scenario A: Automated auditing of viral claims
To check a list of suspicious keywords or entities against global debunks, use the Google FactCheck Claim Search API.
Table 1: API Query Parameters for Investigative Value
Scenario B: Implementing ClaimReview for your investigation
When publishing a debunk, you must “tag” it so automated systems can find it.
Access the Markup Tool: Go to the Google Fact Check Markup Tool.
Enter Claim Details: Input the specific claim, the author of the claim (e.g., a specific politician), and your rating (False, Misleading, etc.).
Embed JSON-LD: Copy the generated code into the
<head>of your article.
💾 3.3. Data preservation and chain of custody
Automated results disappear or change. You must:
Archive the API Response: Save the raw JSON output of your API query.
Snapshot the Source: Use Hunchly or Archive.today to capture the original claim and the debunking article.
Hash the Evidence: Generate a SHA-256 hash for any images or documents you are debunking.
🧠 4. Verification and analysis for reporting
4.1. Corroboration strategy
Never rely on a single automated debunk. Cross-reference an API finding with:
Primary-source audit: If the API says a video is from 2018, find the original from 2018.
Metadata verification: Use InVID/WeVerify to check if the video’s upload date aligns with the fact-check.
4.2. Linking data to narrative
🤖 4.3. AI assistance in analysis
Journalists can use LLMs to process large volumes of fact-check data:
Clustering: Upload a CSV of 500 API results and ask the AI to “Identify the top 3 recurring narratives and the most frequent sources of these claims.”
Summarization: Use AI to summarize foreign-language debunks found via the API.
⚠️ Warning: AI can hallucinate fact-check ratings. Always click the source link provided in the API response to verify the actual rating. Do not upload confidential source documents or whistleblower data to public AI models as this data is used for training and voids source protection.
🚀 5. Practice and resources
5.1. Practice exercise
Go to the Fact Check Explorer.
Search for a current viral topic (e.g., “Climate change 15 minute cities”).
Identify a debunk from a country other than your own.
Find the original “Claim Appearance” URL mentioned in that debunk and use a reverse image search to see if it is still active on other platforms.
5.2. Advanced resources
Eurovision News Spotlight: A collaborative EBU network providing forensic OSINT analysis, Field Notes, and technical tutorials for public service media, investigative journalists and the public.
Google Fact Check API Docs: Full technical documentation for developers.
Schema.org ClaimReview: The official technical definition of the markup.
Duke Reporters’ Lab: A global database of active fact-checking organizations.
✅ 6. Key takeaways and investigative principles
Automate early: Run API queries at the start of an investigation to avoid redundant effort.
Trust verified networks: Leverage the expertise of the Eurovision News Spotlight network for complex cross-border claims.
Machine-readability is key: Use ClaimReview so your hard work is indexed by algorithms.
Context is king: An automated rating doesn’t explain the “why”.
Human-in-the-loop: Always fact-check the automated fact-checkers.
👁️ Coming next week…
Advanced GEOINT: Satellite imagery & temporal analysis
Master the forensic fusion of Google Earth, Sentinel Hub, and OSM to detect, map, and verify environmental and infrastructural change over time.