Companies fear that their eagerness to adopt AI has made them more vulnerable to cyber-states. Recent survey Out of more than 350 IT leaders, more than half of the managers interviewed said that the complexity of AI requests had weakened the cybersecurity posture of their organization. More than two fifths of leaders believe that their security teams do not have the skills necessary to protect their applications and workloads in AI, according to another survey.
The demand for solutions to secure AI applications has generated an entire cohort of startups. Hiddenlayer and Protect IA, for example, focus on the defense of AI systems against contradictory attacks, while Cranium offers visibility in AI systems at the level of the application.
A new company emerging from stealth today, Noma securityDo all this and more, says the co-founder and CEO Niv Braun. The startup is developing tools to identify vulnerable data pipelines and code in data science environments, as well as threats such as rapid injection attacks.
Braun launched Noma with Alon Tron after leaving the Automation Company of the Verint Customer Experience, where he noticed a “significant dead angle” for security teams in the life cycle of data and AI.
“Existing tools are designed for the life cycle of traditional software,” said Braun. “However, the life cycle of data and AI is intrinsically different. It involves distinct R&D processes, relies on different technologies and faces unique technical vulnerabilities in AI models. »»
Noma is designed to help identify – and correct – configuration errors in the components used to develop and run AI applications. For example, the platform analyzes “sensitive” data in model training data sets, such as text containing personally identifiable information.
Braun maintains that companies can be better equipped to deal with threats if they can consolidate AI applications safety functions in a dashboard that can be deployed in cloud-based and self-hosted environments.
“Many unique functional safety solutions flood the market, trying to capitalize on AI of media threw,” he said. “By helping companies understand their imprint of application of AI and giving them governance controls, the noma helps not only organizations to minimize and mitigate risks, but also to give them the confidence they have Need to extend their use of AI faster and to higher parts of the company of the company. “”
Noma, founded in 2023, already has paid customers, says Braun, including fortune customers 500 in B2B software, financial services and retail. The startup recently closed a series of funding of 25 million dollars which, according to Braun, was pre -empted by investors; Noma had money to lose.
“We have decided to accept ($ 25 million), even with substantial funds still to the bank, to accelerate our development and in particular our teams and our entry into service,” he said, Adding that Noma plans to double the size of her team of 20 people in the coming year.
Ballistic Ventures led the Noma series A series, which brings the company’s total capital to $ 32 million. Noma has raised $ 7 million in a series of seeds not disclosed before led by Glilot Capital Partners with the participation of Cyber Club London.
Techcrunch has a newsletter focused on AI! Register here To get it in your reception box every Wednesday.