Peter Thiel-backed artificial intelligence development startup Sentient is looking to differentiate itself in terms of security with the launch of a new system for deploying AI applications in Trusted Execution Environments.

The new Sentient Enclaves Framework v0.70 brings the concept of “confidential computing” to AI development. Announced today, it’s meant to ensure full data isolation, verifiability and attestation for AI applications, the company said. It uses Amazon Web Services Inc.’s AWS Nitro Enclaves technology to ensure that neither AWA nor the host system is able to access or modify AI workloads. In that way, it says, it provides rock-solid guarantees around AI data security.

TEEs create a hardware-enforced security boundary directly on the chip that processes the data used by AI models to prevent unauthorized access by anyone, including the owner of the server. According to Sentient, it paves the way for enterprises to build and run AI applications powered by their most sensitive data, with iron-clad guarantees that this information will never be exposed or compromised.

As a result, AI agents can safely access extremely sensitive details such as people’s credit card information, personal messages and calendars, which can power more personalized AI applications, the company said.

The Sentient Enclaves use AWS Nitro as a foundation to ensure that applications run as intended, without any possibility of nefarious actors making unauthorized modifications. They’re fully open source too, meaning they’re accessible to anyone who’s interested in using them.

Open source is the name of the game for Sentient, which raised $85 million from backers including Peter Thiel’s Founders Fund last June to try to break the dominance of centralized AI model developers such as OpenAI and Google by encouraging collaborative community development.

With Sentient’s platform, developers can work together on the development of open-source large language models that rely on shared datasets and decentralized computing resources. Its platform is built on blockchain technology, and its ecosystem uses cryptocurrency to reward participants based on their contributions.

Sentient co-founder Himanshu Tyagi said the Sentient Enclaves Framework will enable AI developers to “engrain loyalty” in their AI models. “Our user-friendly TEE framework enables Loyal AI — making sure users can trust the code running their AI applications,” he said.

According to Tyagi, loyalty in AI requires the implementation of “technical guarantees” that such systems are running as promised, without any possibility of tampering or surveillance. He believes TEEs are the best available technology to do that.

“Our framework makes one of the most powerful security technologies accessible to anyone,” he said. “Developers can safely build AI applications that protect user data from potential attackers, as well as service providers and cloud operators.

The framework is available now, and in the coming weeks Sentient will introduce an update – the v0.7.1 release – that introduces file system attestation capabilities to enable real-time verification of any job, file system change or execution process, so users will be able to trust the entire runtime behavior of its secure AI models.

Sentient says TEEs are needed because there is growing concern in the industry about how AI systems handle user’s private data. It claims that the framework will be able to address questions raised by the Irish Data Protection Commission’s investigation into X Corp.’s use of EU citizen’s data to train its Grok AI models, for example.

Cryptocurrency exchanges including Coinbase Global Inc. and Biconomy Ltd. have already deployed secure AI applications within Sentient’s Enclaves Framework so they can leverage sensitive customer data, the company said. It didn’t specify the nature of these applications.

In the future, Sentient wants to go beyond AWS Nitro and support additional TEE environments such as Intel Corp.’s Trust Domain Extensions, Advanced Micro Devices Inc.’s Secure Encrypted Virtualization and Nvidia Corp.’s Multi-Instance GPU technology.

Image: SiliconANGLE/Dreamina

Your vote of support is important to us and it helps us keep the content FREE.
One click below supports our mission to provide free, deep, and relevant content.  
Join our community on YouTube
Join the community that includes more than 15,000 #CubeAlumni experts, including Amazon.com CEO Andy Jassy, Dell Technologies founder and CEO Michael Dell, Intel CEO Pat Gelsinger, and many more luminaries and experts.

“TheCUBE is an important partner to the industry. You guys really are a part of our events and we really appreciate you coming and I know people appreciate the content you create as well” – Andy Jassy

THANK YOU