As companies move forward with leveraging their proprietary data in generative AI applications, enterprises are contending with existing security solutions that may be inadequate for that task. Israeli startup Noma Security is addressing that concern. Just out of stealth mode, Noma has raised $32 million in a Series A round led by Ballistic Ventures with support from Glilot Capital Partners, Cyber Club London and a collection of angel investors. While enterprise firms that host their models at large cloud outfits have access to built-in MLOps security tools, those who are self-hosting, using smaller cloud operations, or want added protection might be interested in Noma.

The data and AI lifecycle “is significantly different from the software development lifecycle, with a whole new supply chain, as well as unique open source components and runtime artifacts that traditional security tools don’t cover,” Noma Security co-founder and CEO Niv Braun said in an announcement.

Misconfigured data pipelines and machine learning and DevOps (MLOps) security tools can lead to vulnerable and even malicious models subject to attacks like code poisoning and runtime prompt injection, particularly when using unverified open-source models, according to Braun.

“The demand for solutions to secure AI apps has spawned an entire cohort of startups,” writes TechCrunch, explaining “HiddenLayer and Protect AI, for example, focus on defending AI systems from adversarial attacks, while Cranium provides visibility into AI systems at the application level.”

“Noma provides a comprehensive security platform that ensures the integrity of enterprise customer’s data from the very start, before they do anything to it, all the way through to leveraging it to train and/or deploy AI models and custom applications,” VentureBeat notes, adding that “the platform is already in use by several Fortune 500 companies.”

Braun and Noma co-founder and CTO Alon Tron provide additional details in a blog post about Noma security.

CIO Dive reports on a recent study by IT solutions firm Flexential that found more than half of those surveyed “credit the complexity of AI applications for weakening their company’s cybersecurity posture by expanding the attack surface,” while “around 2 in 5 IT leaders say their security teams lack the skills needed to protect AI applications and workloads.”