Skip to content
Model IP protection

Protect your AI model

Your AI model stays yours. Protect your model. Wherever it runs: on partner and customer servers, in their cloud, on your own servers and plants, and across your edge fleet, lines and field devices.

Where models go

Four places your model goes

Your model earns money wherever it runs. Wherever someone can reach it, it can be copied.

  1. 01

    Partner and processor servers

    Risk: An administrator or contractor at the partner can copy the model file or read it from memory.

  2. 02

    Customer servers and cloud

    Risk: The customer keeps a working copy after the contract ends, or hands it to someone else.

  3. 03

    Your own servers and plants

    Risk: Insiders, contractors and service staff can reach the machines the model runs on.

  4. 04

    Edge fleet, lines and field devices

    Risk: Devices are serviced, resold, lost or recovered, and the model can travel with the hardware.

Four independent layers, modular defence in depth

Wodan protects the model itself

How it works in detail

Explore each ring

Layer 01

Locked

The model file is encrypted and its key is tied to the device it is deployed to, so a copied file is built not to run elsewhere.

Solutions

Where this matters first

Where we are working

Conversations under way in four sectors

We are working with companies in these sectors. We do not name them.

  • Food manufacturing
  • Automotive supply chain
  • Power grid operation
  • Defence supply chain
Why this matters

On-device models are often easy to lift.

A 2021 USENIX Security study by Sun and colleagues examined 1,468 Android apps with on-device machine learning. 41% of them did not protect their models at all. Of those that did protect it, 66% were still extracted with simple tools.

Sun et al., Mind Your Weight(s), USENIX Security 2021
Threat model

How the options compare

See the full comparison
  • Wodan layered protection

    Where the model runs
    Partner, customer and own servers, cloud and edge devices
    Model file on the host
    Locked, so it is built not to run elsewhere
    While the model runs
    Shielded at runtime, with weights hidden inside the model
    Who can read results
    Encrypted results for authorised readers, where enabled
  • Licence contract

    Where the model runs
    Wherever the licensee installs it
    Model file on the host
    Readable once delivered
    While the model runs
    Relies on the licensee keeping terms
    Who can read results
    The licensee
  • Own cloud or API

    Where the model runs
    Only in your own cloud
    Model file on the host
    Never delivered
    While the model runs
    Stays with you
    Who can read results
    Whoever calls your API
  • Hardware enclave

    Where the model runs
    Inside the protected area of specific supported chips
    Model file on the host
    Encrypted outside the protected area
    While the model runs
    Protection depends on the chip vendor
    Who can read results
    Depends on the application
Insights

Why model IP is on the table now

All insights

Bring one model and one deployment.

We map which layers fit your hardware and risks, and what a paid pilot would confirm.

Our engineers review your model and deployment context with you.