NexAI Privacy, Data Safety & Terms Portal

NexAI (package com.nexai.nexai), published by Bengal Bytes, is architected from the ground up as a local-first client. Unlike cloud-hosted chat platforms, Bengal Bytes does not require user registration and operates zero backend chat servers.

Every conversation thread, custom system persona, local document attachment, and preference setting is stored strictly inside your Android device's private application SQLite sandbox.

• Zero mandatory email, phone number, or social login required.
• Personal Bring-Your-Own-Key (BYOK) tokens are encrypted locally via Android Hardware Keystore (AES-256-GCM).
• Voice input (when invoked by tapping the microphone icon) streams strictly while active and is never recorded in the background.

When you choose a cloud AI provider (such as OpenRouter, Google Gemini, Groq, or DeepSeek), your prompt payload is transmitted directly over TLS 1.3 to that selected provider under their respective API data policy.

To sustain free development without mandatory subscriptions, NexAI displays standard banners/interstitials via Google AdMob (with Google UMP GDPR/EEA consent controls) and collects anonymous crash stack traces to fix stability issues.

Because NexAI creates no server-side account on Bengal Bytes infrastructure, you hold 100% immediate self-service deletion power directly on your phone:

• Wipe All Chats & Keys In-App: Open NexAI → Settings → Privacy & Storage → Clear All Chats & API Vault.
• Android System Wipe: Go to Android Settings → Apps → NexAI → Storage → Clear Storage, or uninstall the app to permanently destroy the encrypted SQLite database and Keystore entries.
• Ad ID Reset & Support Inquiries: Reset your Google Advertising ID anytime in Android Privacy Settings, or contact [email protected] for human assistance within 24 hours.

By installing or using NexAI, you agree not to attempt to bypass safety filters or use third-party or local models to generate unlawful, abusive, deceptive, or harmful material. NexAI is rated 13+.

Large Language Models generate probabilistic text and code that may contain factual inaccuracies ('hallucinations'). Always independently verify code, legal, financial, or medical information before acting on it.