How VerifyAI actually works under the hood

People keep asking how my AI fact-checking extension calculates those reliability scores. Here's the breakdown:

The process is pretty straightforward:

When ChatGPT says "France has 68.6 million people," my extension automatically searches Wikipedia, government databases, and Google for that exact claim.

If Wikipedia says 68.6M and INSEE (French stats office) confirms it boom, 92% reliable score. If sources contradict each other or I find nothing - low score.

The scoring:

- 3+ sources agree = 90%+ reliable

- 2 sources agree = 70-80%

- Sources disagree = 30-60%

- No sources found = under 30%

Everything's clickable so you can verify my verification. The whole point is transparency.

Where it breaks down:

Works great for basic facts, not so much for opinions or super niche topics. And yeah, it's only as good as the sources it finds.

Had someone ask if it catches deepfake transcripts - nope, that's way beyond what this does. It's just cross-referencing factual claims, not detecting AI-generated content.

Anyone building similar verification tools? Curious what approaches others are taking.