Israeli Cybersecurity Startup Corma Raises $60 Million to Advance AI Defense Models
Corma, an Israeli startup specializing in language models for defensive cybersecurity, has secured $60 million in a seed funding round led by Sequoia Capital, with participation from Khosla Ventures and Coatue. Founded last year and based in Tel Aviv with 20 employees, the company is led by CEO and founder Alon Floda, who previously co-founded and served as CTO of Migo Security before departing and settling a legal dispute with the firm.
Floda explained that Corma trains proprietary language models specifically designed for cybersecurity defense, which outperform other models in this domain. The company integrates these models as virtual security employees within organizations, complementing existing security tools and personnel rather than replacing them. This approach allows companies to scale their cybersecurity workforce virtually, with the AI models performing a wide range of defensive tasks.
Recent advances in AI, such as ChatGPT and Gemini, have enhanced capabilities in language processing and code generation, which can be exploited by cyber attackers to develop sophisticated attacks. However, defensive cybersecurity requires processing vast amounts of logs, network traffic, and security events over time, a challenge for general AI models. Corma conducted internal simulations mimicking Fortune 500 environments, where AI attackers succeeded 88% of the time, while AI defenders detected only 12% of threats, highlighting a critical vulnerability in current AI defense.
Corma’s model underpins its "General Security Employees," designed to continuously learn and enhance organizational security teams’ effectiveness. The company envisions this as a superhuman defensive workforce capable of rapid, large-scale, and sophisticated responses to the growing AI-driven cyber threats. The seed round was completed in early 2026, and Corma is collaborating with numerous leading companies, with no immediate plans for another funding round.
