Privacy Policy & Data Security
At Jutt AI Hub and Jutt Cyber Tech™, we believe cybersecurity starts with fundamental user privacy. We do not record, store, profile, or monetize your search queries, prompts, models, or private workloads.
Zero Recording of User Prompts
We never capture, log, store, or monitor your search inputs, model prompts, inference queries, or local AI commands. What you search and run remains strictly on your device.
Direct Hugging Face API Fetching
Jutt AI Hub connects directly to the public Hugging Face API and CDN to fetch open-weights, tags, and model metadata in real-time without passing through intermediate logging proxies.
100% Local Inference & Offline Safe
Models executed via Ollama, PyTorch, Diffusers, GGUF, or llama.cpp run locally on your GPU/CPU hardware. No model outputs or proprietary data are ever transmitted to Jutt Cyber Tech.
Safetensors Security & Defense
We actively advocate for Safetensors cryptographic weight verification and isolated sandboxing to protect developers from arbitrary code execution and malicious deserialization vectors.
01 Information We Do NOT Collect (Zero-Telemetry Commitment)
Jutt Cyber Tech™ operates with a cybersecurity-first principle of data minimization and zero unnecessary telemetry. When you use Jutt AI Hub (ai.juttcybertech.com):
- No Prompt or Workload Storage: We do NOT intercept, store, record, or inspect any natural language prompts, text queries, image inputs, code snippets, or audio inputs you process through AI models.
- No Model Search Profiling: We do NOT build behavioral user profiles based on the categories, models, or tags you search for or filter.
- No Model Weights Transmission: When you download or pull weights using commands (e.g.
ollama runorhuggingface-cli download), files flow directly between your machine and official repositories without routing through our private servers. - No User Training Datasets: We do not capture or utilize any user conversations, synthetic data, or fine-tuning datasets to train proprietary AI models.
- No Hardware Fingerprinting: Dynamic VRAM and GPU requirement estimates are computed entirely in your local browser JavaScript engine without transmitting your hardware configuration to our servers.
02 Hugging Face API Integration & External Repositories
Jutt AI Hub functions as an advanced interactive portal that indexes, organizes, and discovers open-source machine learning models from Hugging Face (huggingface.co).
To provide real-time updates on trending models, downloads, likes, model sizes, and pipeline tags, our front-end interface communicates with public endpoints provided by Hugging Face. When interacting with these external resources:
- Your browser may make direct HTTPS requests to
huggingface.co/api/modelsto retrieve live model metadata and cards. - Any interaction with Hugging Face's official infrastructure is governed by Hugging Face's Privacy Policy and Terms of Service.
- Jutt Cyber Tech™ does not inject trackers, credentials interceptors, or man-in-the-middle proxies into Hugging Face API transmissions.
03 Client-Side Processing & Local Execution Security
Security and confidentiality are core pillars of Jutt Cyber Tech™. All interactive tools featured on Jutt AI Hub—including the dynamic VRAM Calculator, Command Generation Engine, Safetensors Format Checker, and Model Comparison Matrices—execute 100% locally in your client's web browser memory.
When running models locally using runtimes like Ollama, llama.cpp, vLLM, PyTorch, or Hugging Face Diffusers:
- Execution occurs strictly within your local GPU/CPU compute boundaries and operating system memory.
- No telemetry, generated completions, or intermediate activations are sent back to Jutt AI Hub.
- You maintain full sovereign control over your enterprise IP, proprietary prompts, and confidential data.
04 Cookies, Local Storage & Advertising Disclosures
To maintain transparency regarding standard web technologies:
- Local Storage: We use your browser's
localStoragesolely to remember client-side preferences (such as selected theme modes, active filters, or dismissed notice banners). This data never leaves your device. - Google AdSense: To support independent cybersecurity research and infrastructure upkeep, we may display non-intrusive advertisements served by Google AdSense (
ca-pub-1101940996678668). Google may use cookies or web beacons to serve ads based on prior visits to this or other websites. You can manage personalized advertising preferences via Google Ads Settings or through the Digital Advertising Alliance. - No Third-Party Data Brokers: We do NOT sell, rent, trade, or distribute your IP address, email, or usage data to third-party data brokers or marketing firms.
05 Model Weights Safety & Deserialization Defense
As a professional cybersecurity consulting firm, Jutt Cyber Tech™ actively educates users about supply chain threats in machine learning artifacts.
Older PyTorch model weights serialized with Python's pickle format (.bin, .pt, .pkl) can inherently harbor arbitrary code execution payloads upon unpickling. Jutt AI Hub strongly encourages users to:
- Prioritize verified Safetensors (
.safetensors) and GGUF (.gguf) formats that do not permit arbitrary executable bytecode during memory mapping. - Inspect SHA-256 cryptographic hashes before loading weights into privileged enterprise environments.
- Run untrusted third-party community models inside sandboxed containers or isolated virtual environments with restricted outbound network privileges.
06 Global Privacy Rights & Compliance (GDPR, CCPA)
Because Jutt AI Hub does not collect personal identifiers, account credentials, or behavioral logs, our data retention footprint for public visitors is virtually zero. Nevertheless, in compliance with GDPR, CCPA/CPRA, and international standards, you hold the right to:
- Inquire about any potential data processed if you contact us directly via email or communication channels.
- Request immediate deletion or correction of any correspondence or consulting inquiries you transmit to us.
- Opt out of advertising cookies through your browser settings or privacy extensions.
07 Data Controller & Contact Information
If you have questions regarding this Privacy Policy, our zero-telemetry architecture, or need security consulting for your AI deployments, please reach out directly: