Confidential AI
Train AI models on end-to-end encrypted data
With data privacy regulation evolving worldwide, companies have to ensure AI data compliance. With confidential computing, you can keep machine learning training data secure and always encrypted - even during processing.
Isolating AI pipelines end-to-end for data privacy compliance
When training AI models, companies collect a lot of personal data that needs to stay secure, especially in public cloud environment. With Confidential AI, ML data stays encrypted end-to-end, even while in use. Using Constellation, companies process ML data securely on public cloud infrastructure with verified encryption.
Why data science & compliance teams love confidential AI
Building a GDPR-compliant Confidential AI pipeline for Bosch
Bosch works at the forefront of artificial intelligence and autonomous driving. As data privacy and compliance are major concerns when gathering and training neural networks, Bosch wanted to leverage confidential AI. Based on EGo and EdgelessDB, we have built an always encrypted, confidential AI pipeline. With the solution, Bosch strengthened AI data compliance and security.
Please accept marketing-cookies to watch this video.
Confidential AI solves data privacy and compliance for good
With Confidential AI, you can keep data always encrypted, even while in use. Confidential AI enables companies to train neural networks on encrypted data, in line with privacy laws and without the need for anonymization or tokenization.
With our enterprise product Constellation, AI workloads can be run in isolated clusters on public infrastructure. Our product enables lifting & shifting existing AI workloads securely. Confidential AI is the easiest way to become compliant with data privacy at scale.
Turn the public cloud into your private cloud.
Confidential AI soon also available for GPU-based workloads
While confidential computing hardware is currently available only for CPUs, NVIDIA has recently announced GPU processors that enable runtime encryption for AI workload security. Once the hardware is there, we will make sure to offer Confidential AI software also for GPU-based workloads. Confidential AI will enable a new era of trustworthy AI data processing. Let’s make AI secure and GDPR-compliant with true end-to-end confidentialilty.
In this video you can listen to Mark Overby (Chief Architect, NVIDIA) talk about the Hopper GPU attestation architecture which is used to measure the necessary security posture to protect your valuable data in-use.
Please accept marketing-cookies to watch this video.
Any Questions?
Interested in learning more about Confidential AI? Contact us!