{
  "id": 20351533,
  "title": "Corma raises $125M to train cybersecurity-optimized foundation models",
  "url": "https://urgent.news/2026/11/16/corma-raises-125m-to-train-cybersecurity-optimized-foundation-models",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-11-16T23:48:10.000Z",
  "source": {
    "name": "SiliconANGLE",
    "slug": "siliconangle",
    "url": "https://siliconangle.com/2026/11/16/corma-raises-125m-to-train-cybersecurity-optimized-foundation-models/"
  },
  "original_language": "en",
  "account": "Cybersecurity-focused AI startup Corma.ai Ltd. has secured $125 million in funding to develop models capable of detecting and resolving cyber threats. The recent Series A round was spearheaded by NEA, with additional contributions from Khosla Ventures, Coatue, and Sequoia Capital. The company's valuation now stands at $1 billion.\n\nCorma's office locations include San Francisco and Tel Aviv. The company claims that competing AI models often excel at creating cyber attacks rather than preventing them. To demonstrate this, Corma conducted a test involving multiple large language models infiltrating a simulated corporate network. They discovered that these models successfully installed malicious files in 85% of the test runs, while only identifying 12% of the malicious files they themselves had created.\n\nThe discrepancy between offensive and defensive capabilities of Large Language Models (LLMs) presents an advantage for cybercriminals over cybersecurity teams. Corma aims to address this issue by creating cybersecurity-optimized AI models. These algorithms can expedite breach response workflows by 100 times and decrease associated costs. Corma benchmarked its model against competitors using DefenseBench V5, a measurement that evaluates a model's reliability and speed in defensive cybersecurity tasks.\n\nCorma's algorithms locate security breaches by examining data from various cybersecurity tools, such as firewalls and network monitoring software. The company's models not only identify threats but also provide remediation guidance for each one. Automating these breach response tasks is challenging due to the need for LLMs to filter through numerous false positives and duplicate alerts. Additionally, models often receive incomplete information about breaches, as many companies delete cybersecurity logs after a short period to cut data storage expenses.\n\nDespite these challenges, Corma's algorithms can perform autonomous breach investigations. If human input is required, the system alerts a relevant cybersecurity team member. Alternatively, an administrator can manually control the models via a Slack chatbot. For instance, they could ask Corma to verify if a recent malware removal from a compromised server was successful.\n\nSince leaving stealth mode, Corma has faced overwhelming demand from leading organizations, making it difficult to maintain sufficient capacity. CEO Alon Pluda stated that the newly raised funds will be allocated toward acquiring more infrastructure and further enhancing the company's AI models.",
  "summary": "Corma.ai Ltd., a startup that develops artificial intelligence models for cybersecurity teams, has raised $125 million in funding. NEA led the Series A round. Corma stated today that returning backers Khosla Ventures, Coatue and Sequoia Capital chipped in as well. The company is now valued at $1 billion. Corma, which has offices in San Francisco […] The post Corma raises $125M to train…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}