With Privatemode, you can finally use generative AI without having to worry about data privacy and security.
Privatemode is an AI service like ChatGPT—but with one critical difference: your data stays private.
Using confidential computing, Privatemode encrypts your data before it leaves your device and keeps it protected even during AI processing. This ensures that your information remains secure at all times.
Get a secure AI assistance for your daily work. Generate content and analyze documents while keeping business data private.
docker run -p 8080:8080 \
ghcr.io/edgelesssys/continuum/continuum-proxy:latest \
--apiKey <your-api-token>
curl localhost:8080/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "ibnzterrell/Meta-Llama-3.3-70B-Instruct-AWQ-INT4",
"messages": [
{
"role": "system",
"content": "Hello Privatemode!"
}
]
}'Get on-premises like privacy combined with the flexibility and scalability of the cloud.
All you need is an API key and our proxy, which automatically encrypts your data and verifies the integrity of the backend.
Privatemode lets you use the latest open-source AI models even with sensitive data. Privatemode’s architecture prevents the model from leaking your data and the infrastructure from accessing your data.
With confidential computing, your data remains encrypted at all times - during transfer, storage, and during processing in main memory.
The Privatemode app and proxy verify the integrity of the Privatemode service based on hardware-issued cryptographic certificates.
The Privatemode service is architected to prevent any external party from accessing your data, including even Edgeless Systems.
The Privatemode service is hosted in top-tier data centers in the European Union. More locations are coming soon.
The confidential computing-based architecture of Privatemode ensures that the employed AI model can technically not remember your data and cannot be re-trained on it. The architecture also keeps everyone else out—including us.
Privatemode uses a technology called confidential computing to protect your data at all times—even while it's being processed by the AI.
Privatemode is designed in a way that ensures no one can access your data, including Edgeless Systems, the company behind Privatemode, and the cloud provider that the Privatemode backend runs on.
Privatemode currently uses a "quantized" version of Meta's Llama v3.3 with 70B parameters. The option to choose between models will be added over time.
No. Privatemode's architecture ensures that the employed AI model can never learn from your data or remember it in any form.
Privatemode is built around a technology called confidential computing. With it, data can not only be kept encrypted during storage and transmission, but also during processing in main memory.
The CPU and GPU hardware that is running the Privatemode AI model in the backend is able to selectively decrypt your data into its internal structures. These internal structures are inaccessible from the outside. In fact, from the outside, it looks as if the hardware was processing encrypted data.
Even highly privileged attackers like sysadmins or data-center employees will only ever see encrypted data.
The Privatemode app verifies the integrity of the Privatemode service before sending any data, using a mechanism called remote attestation. The app also encrypts all data before transmitting it to the service. These tasks cannot be performed efficiently within a browser.
Yes, apps for iPhone and Android will launch soon.
