False Social
How False Social works
False Social is a fully automated fact-checker. An agentic workflow built on OpenAI's agent SDK ingests each post, splits it into phrases, and assigns every phrase a classification, truth score, verifiability score, confidence, rationale, and supporting evidence.
Not every post is analyzed. Posts Trump makes on Truth Social are filtered before analysis, including meeting a minimum character-count threshold. This project was retired on July 4th 2026, once Akshad ran out of OpenAI API credits.
Phrases are highlighted on the truth spectrum. The stronger the color, the more verifiable the claim:
- Green: accurate
- Amber: mixed
- Red: inaccurate
A post's overall truth and verifiability scores are weighted sums of its phrases. Hover over or tap any phrase to see the rationale and evidence behind its score.