🛡️ Trust Score Methodology
The problem: researchers at CMU and NC State found ~6 million fake GitHub stars across 18,617 repositories — and AI/LLM projects are the #1 non-malware category for fake stars (~177,000 of them). Stars cost as little as $0.03 each. Most "top AI tools" lists never check.
Every tool in our AI Tool Finder carries a Trust Score (1–10). It doesn't measure how good a tool is — the star rating does that. It measures whether you can believe what the vendor tells you.
Current database stats
- 37 tools verified
- 9 tools scored 8+/10 (high trust)
- 10 tools currently carry active red flags
- Average Trust Score: 6.6/10
The 5 scoring dimensions
1. GitHub metric authenticity (0–2 points)
For open-source or GitHub-hosted tools, we sample stargazer accounts. Research shows36–76% of fake stargazers have zero followers, no repos, and account creation dates clustered within days. Closed-source tools are marked not_applicable and scored on the remaining dimensions.
2. Pricing transparency (0–2 points)
Published pricing = full marks. "Book a demo" walls, hidden usage fees, or bait-and-switch trial terms lose points. If we can't find out what it costs without a sales call, users can't either.
3. Review authenticity (0–2 points)
We look at review distribution patterns on G2, Trustpilot, and Product Hunt: bursts of 5-star reviews in short windows, reviewer account age, and template-like phrasing. The FTC now fines fake social proof at $53,088 per violation — but enforcement lags far behind reality.
4. Company track record (0–2 points)
Identifiable team, real company registration, funding history, and product longevity. Anonymous teams with 30-day-old domains selling lifetime deals score zero here.
5. Marketing accuracy (0–2 points)
We compare landing-page claims against documented capabilities. "Replaces your entire team" with no benchmark, no case study, and no free trial is a claim we penalize.
What the scores mean
| Score | Label | Interpretation |
|---|---|---|
| 8–10 | 🟢 High trust | Claims verified, transparent pricing, authentic metrics |
| 6–7 | 🟡 Moderate trust | Legitimate product, some claims unverifiable or pricing opaque |
| 1–5 | 🔴 Caution | Active red flags — verify independently before paying |
Red flags
Independent of the score, tools can carry red flags — specific, documented concerns (suspicious star growth, undisclosed data usage, review manipulation). Flags appear directly on tool cards in the Finder and are removed when resolved.
Re-verification
Every tool carries a verification_date. Scores are re-checked when a tool ships major changes, changes pricing, or when readers report discrepancies. The full dataset — including trust fields — is available to AI engines and researchers at/api/tools.json.
How we make money
Some outbound links are affiliate links: if you buy through them, we may earn a commission at no extra cost to you. Two hard rules keep this honest: trust scores are never affected by affiliate status — tools with no affiliate program are scored identically to tools that pay — and red flags are never removed for commercial reasons. Rankings on this site cannot be bought. If you ever spot a conflict between our data and our links, email[email protected] and we'll investigate publicly.
Find AI tools you can actually trust
Every tool scored across 5 trust dimensions. No pay-to-play rankings.
Open the AI Tool Finder →Sources
- "4.5 Million (Suspected) Fake Stars in GitHub" — CMU / NC State / Socket, ICSE 2026
- FTC Rule on Fake Reviews and Testimonials (16 CFR Part 465), effective October 2024