In recent years, the rapid advancement of AI and machine learning technologies has enabled the growth of transformative applications across industries. However, the deployment of these systems has unveiled critical challenges pertaining to their robustness. To address these issues, organizations need purpose-built tools that help eliminate the vulnerabilities of these machine-learning models, without sacrificing the speed of application modernization. These vulnerabilities not only compromise the performance and security of AI applications but also hinder their widespread adoption in high-stakes domains such as healthcare, autonomous vehicles, and finance. 

Wiz is excited to announce support for Amazon SageMaker, enabling data scientists and machine learning engineers to accelerate their models into production faster and more securely. This support helps secure the ML models that customers build, train, and deploy in SageMaker, including access to SageMaker Notebooks, and more. Wiz provides joint customers full visibility into their cloud technology usage, minimizing their security risks such as external exposure, data protection, and entitlement, so practitioners can focus on innovative code to build AI-powered applications. 

SageMaker empowers machine learning practitioners to build, train, and deploy ML models. Wiz’s support for SageMaker helps protect those models, allowing our joint customers to innovate faster and more securely in AWS.

Michele Monclova, Senior Manager of Product Management, Amazon SageMaker 

Customers that use SageMaker Notebook instances now can have peace of mind because Wiz can monitor and manage the security risks associated with building AI/ML models. For example, it can identify whether that Notebook instance has disabled internet access or when the notebook is publicly accessible. When actors outside the VPC (Virtual Private Cloud) can establish a connection to the notebook, this increases the attack surface and the opportunity for malicious activity. Wiz Cloud Detection and Response (CDR) can also detect pre-signed URLs for SageMaker Notebooks, which would enable malicious actors to potentially gain access without any authentication. 

In addition, Wiz can help identify when a Notebook instance has excessive permissions to sensitive data. Using the Wiz Security Graph, our attack path visualization shows how an external organization user could gain access to models based on sensitive data, which would result in data leakage. We also help protect SageMaker customers from data poisoning, which prevents attackers from injecting incorrect or biased data into the Notebook instance where the machine learning model is built. Successful data poisoning cases can have severe impacts. For example, they can cause models to misclassify inputs or reduce their performance, which would be particularly concerning in critical applications across highly regulated industries. 

One company that is pulling this together is Stability AI. They use Wiz to help empower data scientists and engineers to develop innovative open AI models for image, language, audio, video, and more. 

Stability AI empowers creators, developers, and researchers to pioneer transformative AI models that unlock human potential. Wiz safeguards our infrastructure and APIs to facilitate seamless integration for partners and service providers, all while advancing the frontiers by supporting our needs related to AI research.

Garry Jean-Baptiste, CISO, Stability AI

We are not stopping there. AWS recently announced the launch of Amazon Bedrock, a fully managed service that makes foundation models (FMs) from Amazon and leading AI startups available through an API. Wiz will also help empower practitioners to accelerate their machine-learning journey by protecting their generative AI applications in the real world. Stay tuned for more. 

Enable more practitioners to innovate with machine learning 

Our goal is to help your organization’s data scientists and engineers build AI-powered applications faster, while we protect your machine learning models from leakage and poisoning of sensitive data. You can learn more by visiting our Wiz docs (login required). If you prefer a live demo, we would love to connect with you.