"Catch the bias. Before it catches someone."

FairSight detects hidden discrimination in AI decision systems — in seconds, for free, with zero technical knowledge required.

Every day, AI systems make life-changing decisions about who gets a loan, a job, or medical care. These systems learn from historical data — data that reflects decades of human bias. The AI learns the bias. The bias gets automated. Real people get hurt.

Real documented cases:

• Amazon (2018) — AI hiring tool penalized resumes containing the word "women's"
• COMPAS (2016) — Court sentencing algorithm rated Black defendants 2x as likely to reoffend
• US Hospitals (2019) — Healthcare AI gave less care priority to Black patients
• Indian Banks (2023) — Credit scoring AI systematically rejected rural applicants

Upload any CSV dataset → get a fairness score in seconds → download a compliance PDF.

• Fairness Score 0-100 — Overall bias rating with color-coded risk level
• 4 Fairness Metrics — Demographic parity, statistical parity ratio, 4/5ths rule, high-risk groups
• Explainable Drill-Down — Click any demographic group, Gemini explains exactly why they face bias
• 6-Regulation Compliance Mapper — EU AI Act 2024, India DPDP Act 2023, US EEOC, RBI Fair Practices
• Score Improvement Simulator — Toggle fixes on/off and watch the projected score update in real time
• Multi-Dataset Comparison — Upload before and after datasets, see side-by-side improvement
• Google Sheets Integration — Import data directly from Google Sheets, no CSV download needed
• Professional PDF Report — 4-page compliance document with unique audit ID, regulatory citations
• Audit History — All audits saved to Firebase Firestore, track improvement over time
• Intersectional Bias Detection — Detects bias across combined demographic groups

git clone https://github.com/YOUR_USERNAME/fairsight.git
cd fairsight
npm install

Create a .env file: REACT_APP_GEMINI_KEY=your_gemini_api_key REACT_APP_FIREBASE_KEY=your_firebase_key REACT_APP_FIREBASE_AUTH=your_auth_domain REACT_APP_FIREBASE_PROJECT=your_project_id

npm start

FairSight works with any CSV. Try these real-world bias datasets:

All CSV processing happens entirely in the browser. Your data never reaches our servers. FairSight uses Papa Parse for client-side parsing and processes everything in JavaScript memory. Only anonymized audit scores are saved to Firebase (when signed in).

FairSight checks compliance against:

• EU AI Act 2024 — Article 10 (Data governance) and Article 13 (Transparency)
• India DPDP Act 2023 — Section 4 (Lawful processing) and Section 6 (Consent)
• US EEOC Guidelines — 4/5ths rule for employment selection
• RBI Fair Practices Code — Section 3 (Non-discriminatory lending)

Solution Challenge 2026 India — Unbiased AI Decision Track

Powered by Google Gemini API and Firebase