Eclypsium announced new GenAI assessment capabilities for its Supply Chain Security Platform. The new capabilities help secure the fundamental layers of the GenAI tech stack through support for NVIDIA hardware and popular GenAI foundation models.
As demand for GenAI skyrockets, IT infrastructure is changing rapidly to accommodate training and inference requirements. For example, data center operators are rethinking how they design their facilities to accommodate GenAI workloads with new types of power and cooling systems, and accelerated adoption of such technologies as optical interconnects.
As the GenAI tech stack becomes more complex, the IT industry must address security concerns. Eclypsium’s added support for NVIDIA hardware, such as the NVIDIA H100 Tensor Core GPU, means cloud services providers and other data center operators are able to identify vulnerabilities and insecure configurations and detect compromise, including the presence of implants and backdoors.
Meanwhile, the addition of popular GenAI foundation models to Eclypsium’s supply chain intelligence allows organizations to better understand and compare the risk of various closed- and open-source GenAI foundation models.
“Today, GenAI impacts every organization as we learn how it can make people more productive and our products more effective,” said Eclypsium CEO Yuriy Bulygin. “While GenAI is transforming the way we interact with technology and the world, it creates unique security challenges. We need to understand how to securely build, deploy, and use GenAI models and agents, and how to protect the infrastructure they run on. Eclypsium has developed capabilities to ensure integrity and security of HPC servers, GPUs, and other AI compute resources to protect those fundamental layers of the GenAI infrastructure.”
The new capabilities from Eclypsium allow users to continuously monitor and remediate risks in popular NVIDIA hardware used to train GenAI models. As with standard server infrastructure, the systems used to run GenAI workloads are composed of hardware and firmware components, each with complex supply chains.
Threats at this fundamental level are virtually impossible to detect and remediate without visibility. The Eclypsium platform provides the third-party assurance and validation that cloud services providers and other data center operators need to ensure that their GenAI hardware infrastructure can be trusted.
Support for popular GenAI models allows Eclypsium customers to easily compare the vulnerabilities and software characteristics of models used to train GenAI applications, including Claude (Anthropic), ColBERT (Stanford University), Command R+ (Cohere), GPT (Open AI), DBRX (DataBricks), Gemini (Google), Llama (Meta), Mixtral (Mistral AI), RoBERTa (Meta), WizardLM (Microsoft), and Zephyr (Hugging Face). This enhanced supply chain intelligence enables teams building both internal and commercial GenAI applications to make risk-informed decisions when choosing a foundation model.