Overview
What is Authenta?
Authenta is an AI-powered media authenticity platform that detects deepfakes, AI-generated images, and image forgeries with industry-leading precision.
It helps organizations verify the integrity of visual content before publication, storage, or legal processing.
The Authenta On-Prem Edition is designed for enterprises and government agencies that require complete data control.
It can be deployed entirely inside your infrastructure — operating fully offline once installed — ensuring that sensitive media never leaves your network.
Authenta leverages deep neural models trained on large-scale datasets to identify manipulation patterns in both video and image files.
It integrates easily with your internal applications through RabbitMQ or a lightweight REST adapter.
Key Features
- 🔒 Data Sovereignty — All media files, results, and logs stay within your infrastructure.
- 🧠 AI-Based Detection — Specialized models detect deepfakes, synthetic images, and forged regions.
- ⚙️ Flexible Deployment — Delivered as Docker containers with support for CPU and GPU profiles.
- 🚫 Offline Operation — Requires internet only once for image pull; operates fully air-gapped afterward.
- 🧩 Modular Components — RabbitMQ message broker, ML inference runner, and shared storage volume.
- 📈 Scalable by Design — Deploy multiple inference containers for parallel job processing.
- 🔐 No Telemetry — No external communication, tracking, or analytics.
- 🧾 Compliant — Meets on-prem data isolation and enterprise security standards.
Supported Models
| Model | Input Type | Description | Status |
|---|---|---|---|
| DF-1 | Video | Detects deepfake or manipulated faces in video content. | Stable |
| AC-1 | Image | Determines if an image was AI-generated or synthetic. | Stable |
| FD-1 | Image | Detects forged or tampered regions within an image. | Beta |
Each model can be selected dynamically when publishing a task to RabbitMQ.
All inference operations occur locally within your environment — no cloud calls or external processing are involved.
Offline & Secure Operation
Authenta’s architecture is designed for air-gapped, high-security environments:
- All components (RabbitMQ, inference runner, shared storage) run locally as Docker services.
- The system communicates only within your internal network.
- No telemetry, logging, or model updates are sent externally.
- Internet connectivity is required only once — to pull container images from Authenta’s private AWS ECR.
- After setup, the system can function indefinitely offline.
You may safely disable outbound network access or enforce firewall restrictions once setup is complete.