SafeShare is an enterprise privacy firewall and autonomous Safe Clipboard designed for banking, fintech, and enterprise workflows.

Problem: Employees and support agents frequently copy customer tickets, account [ACCOUNT_1], and passwords into AI tools like ChatGPT and Claude to draft replies or debug issues. This accidentally leaks sensitive citizen data—such as Aadhaar cards, PAN [ACCOUNT_2], bank accounts, and passwords—causing severe regulatory violations under India’s DPDP Act 2023 and RBI banking security guidelines.

What SafeShare Does:

Safe Clipboard for Banking: SafeShare runs silently on Windows and monitors the system clipboard. In less than 50 milliseconds, it intercepts sensitive data and replaces it with safe synthetic placeholders before it can ever be pasted into external AI tools.
Complete Indian Document Coverage: It automatically detects Aadhaar [ACCOUNT_3] (even with irregular digit spacing), PAN cards, Passports, Driving Licenses, Voter IDs, Vehicle RC [ACCOUNT_4], GSTIN, TAN, CIN, DIN, EPFO/UAN, ABHA Health IDs, and FIR/Court Case [ACCOUNT_5].
Conversational Financial Sentences: It understands natural sentences and typos. For example, “[CUSTOMER_1] transferred [AMOUNT_1] from his [ACCOUNT_6]” is automatically sanitized to “[CUSTOMER_1] transferred [AMOUNT_1] from his [ACCOUNT_1]”.
Reversible Synthetic Tokens: Instead of ugly blackouts, it uses structured tokens so downstream AI prompts stay 100% functional and coherent.
Local Air-Gapped Security: The original data never leaves the user’s PC. An immutable audit log is saved locally on disk in audit_log.json with a 1-click clipboard restore option and pause/resume control.

How We Built It: We built SafeShare using Python, native Windows Win32 APIs for zero-latency clipboard protection, and engineered the semantic reasoning architecture around Google Gemma 4 (gemma-4-31b-it and gemma-4-26b-a4b-it) to evaluate regulatory compliance and enterprise risk scores.