Corma
Corma trains foundation models for defensive cybersecurity and deploys them as an AI security workforce alongside security teams.
At a Glance
About Corma
Corma is a company training foundation models toward what it calls superintelligence for defensive cybersecurity. Its models are deployed as an AI security workforce that works alongside security teams. Access is requested through the company's site.
What It Is
Corma builds foundation models focused on defense rather than offense. According to the company, defense works with audit logs, network flows, and security events, and requires consistency across everything at once, which it argues differs from the coding and exploitation skills general-purpose AI is optimized for. The models are delivered as an AI workforce that operates with security teams.
Company Claims and Research
Corma's homepage presents a DefenseBench V5 chart in which it states Corma v0.4 leads other named models on defensive task performance and execution speed. It also links to research it describes as independent, covering AI attack and defense on an enterprise range, reporting that frontier models planted a persistent backdoor in 85% of runs while defenders found 19%. These are vendor-published claims.
Current Status
The company's blog states it emerged from stealth and announced a $125M Series A in November 2026. Prospective users can request access via the site, and a separate page describes the AI Workforce.
Community Discussions
Be the first to start a conversation about Corma
Share your experience with Corma, ask questions, or help others learn from your insights.
Pricing
Request Access
Request access to Corma’s AI security workforce. Corma hasn’t published pricing or access terms.
- AI security workforce for defensive security teams
Capabilities
Key Features
- Foundation models for defensive cybersecurity
- AI security workforce deployed alongside security teams
- DefenseBench V5 benchmark results published by the company
- Published research on the AI offense-defense gap
