Companies are increasingly worried that their enthusiasm for adopting AI has made them more susceptible to cyber threats. According to a recent survey of over 350 IT leaders, more than half of the executives indicated that the complexity of AI applications has compromised their organization’s cybersecurity defenses. Additionally, over 40% of executives feel their security teams lack the necessary skills to protect their AI applications and workloads, as revealed in another survey. This rising demand for securing AI applications has led to the emergence of numerous startups. For instance, HiddenLayer and Protect AI specialize in safeguarding AI systems against adversarial attacks, while Cranium offers insights into AI systems at the application level. A new startup launching from stealth mode today, Noma Security, claims to provide a comprehensive solution, according to co-founder and CEO Niv Braun. The company is focused on creating tools that can identify vulnerable data pipelines and problematic code in data science environments, addressing threats such as prompt injection attacks. Braun, along with co-founder Alon Tron, started Noma after their time at customer experience automation company Verint, where Braun observed a “significant blind spot” in security teams concerning the data and AI lifecycle. “Current tools are tailored for the traditional software lifecycle, ” Braun noted.

“However, the data and AI lifecycle is fundamentally different. It includes unique R&D processes, relies on different technologies, and presents specialized technical vulnerabilities within AI models. ” Noma aims to identify and resolve misconfigurations in the components used to build and operate AI applications. For instance, the platform can detect “sensitive” data in model training datasets, such as texts containing personally identifiable information. Braun believes that companies will be better prepared to tackle threats if they can centralize AI application security functions within a single dashboard, suitable for both cloud-based and self-hosted environments. “The market is saturated with numerous single-function security solutions trying to capitalize on the AI trend, ” he stated. “By enabling firms to understand their AI application footprint and providing governance controls, Noma helps organizations minimize risks while fostering the confidence necessary to expand their AI usage more swiftly and in higher-value areas of their business. ” Founded in 2023, Noma already boasts paying customers, including Fortune 500 firms in B2B software, financial services, and retail. The startup has recently completed a $25 million Series A funding round which Braun noted was initiated by eager investors; Noma had adequate funds available. “We chose to accept the [$25 million], despite having significant reserves, to expedite our development and especially enhance our go-to-market teams and strategies, ” he explained, adding that Noma plans to double its current 20-person team in the next year. Ballistic Ventures led Noma’s Series A, bringing the company’s total capitalization to $32 million, following a previously undisclosed $7 million seed round led by Glilot Capital Partners with contributions from Cyber Club London.