TL;DR: Model weights can still be exposed when they are decrypted in inference environments the owner cannot verify or control, according to Fortanix. Confidential computing has to cover the full CPU and GPU boundary, or plaintext memory remains the point of extraction.
Editorial analysis by NHI Mgmt Group, based on content published by Fortanix: “A Technical Guide for Model Builders on Protecting AI Model IP”.
Key questions
Q: Why do administrator privileges still matter for model IP risk?
A: Administrator access matters because conventional infrastructure often allows privileged operators to inspect plaintext memory during inference.
Q: What are the signs that a confidential computing deployment is incomplete?
A: The main warning signs are CPU-only attestation, key release that is not tied to verified measurements, and no clear statement about GPU coverage.
Q: How should security teams protect AI model weights in third-party inference environments?
A: Security teams should require runtime confidentiality controls that keep model weights protected while in use, not just at rest.
Practitioner guidance
- Require GPU attestation in every deployment review Do not accept confidential computing claims that only cover the CPU path.
- Make key release conditional on attestation success Treat attestation as a hard gate, not a monitoring signal.
- Evaluate the full inference trust boundary Map every point where the model is decrypted, handed off, executed or monitored, including host OS, hypervisor and accelerator layers.
Bottom line: Model weights are still exposed if they must be decrypted in a runtime the owner cannot verify or control.
Glossary terms: Confidential Computing, Trusted Execution Environment, Attestation, Model Weight Extraction
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →
Model IP protection has become a runtime governance problem, not a storage problem. The article shows that encryption at rest does not matter if the model must be decrypted for inference in an environment the owner does not control. That shifts the governance question from whether data is encrypted to whether the runtime boundary is provably trustworthy. Practitioners should treat model deployment as a control decision, not a hosting preference.
A few things that frame the scale:
- Only 44% of organisations are currently using a dedicated secrets management system, according to the 2024 State of Secrets Management Survey.
- 73% of vaults are misconfigured, leading to unauthorised access and exposure of sensitive data, according to the Ultimate Guide to NHIs.
A question worth separating out:
Q: What is the difference between protecting model data at rest and in use?
A: At-rest protection secures the model before it is loaded, but it does not stop exposure once the model must be decrypted to run inference. Protection in use is harder and requires the runtime itself to enforce confidentiality, usually through hardware-backed isolation and attestation. For AI IP, that difference is decisive.
👉 Read our full editorial: Confidential computing is becoming the baseline for model IP protection