How Fake News Sniper Works
Our AI-powered platform helps you verify claims, detect misinformation, and stay informed with evidence-based analysis
The Verification Process
- 1
Claim Submission
Enter any claim you want to verify - from news headlines to social media posts or even voice input.
- 2
Evidence Gathering
Our system searches Wikipedia and other reliable sources to find relevant information about your claim.
- 3
AI Analysis
Multiple AI models analyze the claim and evidence to determine if the claim is supported, contradicted, or unverifiable.
- 4
Verdict Generation
A final verdict is created with a confidence score, supporting evidence, and sources.
- 5
User Feedback
Community feedback helps improve our system and identifies areas where the AI may need human guidance.
AI Technology
Claim Verification (BART NLI)
We use the BART-large-mnli model from Facebook AI to perform Natural Language Inference. This model analyzes whether the evidence entails, contradicts, or is neutral towards the claim.
Evidence Extraction (RoBERTa QA)
Our system uses RoBERTa-base-squad2 to extract relevant evidence from source documents. This Question-Answering model finds the most pertinent information for verification.
Voice Transcription (Whisper)
We use OpenAI's Whisper model to convert spoken claims into text, allowing for easy verification of audio or voice-based claims.
Knowledge Sources
Our primary source of evidence is Wikipedia, one of the most comprehensive and monitored knowledge bases. We also incorporate other reliable reference databases when available.
Advanced Features
Multi-model Ensemble
We combine results from multiple AI models for more accurate verdicts and higher confidence.
Contextual Scoring
Our confidence scores take into account evidence quality and source credibility.
Fallback Strategy
If one model fails, we have backup approaches to ensure you always get a response.
Community Feedback
User votes help us improve accuracy and identify challenging cases.
Explainability
We provide transparent reasoning for each verdict in the AI Lab section.
Limitations & Disclaimer
While Fake News Sniper uses advanced AI to verify claims, please be aware of these limitations:
- AI models can make mistakes and have biases
- Verdicts are based on available evidence which may be incomplete
- The system's knowledge cutoff means very recent events may not be accurately verified
- Some complex claims require human expertise and nuanced analysis
- Always cross-check important information with multiple trusted sources
