Show HN: FakeScan – AI-powered fake review detector for Amazon

After Fakespot shut down in July 2025, I built FakeScan as a free alternative. It uses AI (Llama 3.3 70B) to analyze Amazon product reviews and detect fake/manipulated ones.

How it works: paste any Amazon product URL, and it analyzes review patterns, sentiment distribution, reviewer behavior, and linguistic markers to give you a trust score and detailed breakdown.

No signup required - just paste a URL and get results. Built with Next.js, running on Groq for fast inference.

Would love feedback from the HN community on the analysis methodology and UX.

fakescan.site

2 points

crawde

7 hours ago


1 comment

testbyhuman_tor 6 hours ago

Nice Fakespot replacement. For UX feedback: have you tried watching someone who shops on Amazon regularly paste a URL and walk through the results? The trust score breakdown looks information-dense — I'd be curious whether non-technical shoppers actually know what to do with the sentiment distribution and linguistic markers sections, or if they just want a simple "trust / don't trust" answer.

The no-signup approach is smart for adoption. The main risk is probably that people paste a URL, see a wall of analysis, and bounce before understanding the verdict.