Key facts

Founded
2026
HQ
United States
Employees
—
Funding stage
—
Further facts
  • Timeline: Common Defense launched publicly on 28 September 2026 as an AI-native cybersecurity company built by Quantstamp, which has secured more than US$500bn in digital assets across 1,300+ audit engagements.
  • Entity: The company trades as Common Defense AI Security, Inc.
  • Scale: By 10 November 2026 the Communication Firewall had screened 4,400,352 messages and caught 84,776 threats.
  • Accuracy: In internal testing the system detected 96.7% of scams with a 0.48% false-positive rate and typically decided in under 100 milliseconds.
  • Team: Its researchers come from Quantstamp, ByteDance, Stanford and red-team work involving OpenAI and Anthropic, and have contributed to the OWASP LLM Top 10 and PentestGPT.

Focus areas

Email security Scam detection Communications surveillance Deepfake detection Incident response Digital assets

About

Common Defense is an AI security company, launched in September 2026 by Quantstamp founder Richard Ma and a team from the crypto security firm, that protects people and organisations from social engineering. Its Communication Firewall connects to WhatsApp, Telegram, Slack, SMS, Gmail and Outlook, monitors incoming messages and warns users inside the app when something malicious arrives, while security teams manage policy from a dashboard. It works mainly with AI companies, fintech firms and crypto protocols, where key compromise, signer manipulation and deepfake calls cause large losses. It also offers incident response, operational security reviews, monitoring, certification and virtual CISO services.

How they use AI

Common Defense screens each message through a staged pipeline. Identity and signature checks clear most traffic in milliseconds, a trained detection model handles most of the rest, and only about 3% of messages reach a large language model for deeper judgement. The system looks across channels at once, because the strongest signs of social engineering often appear only when weak signals from email, chat and calls combine. Every attack it meets feeds back into the models. Administrators never see users' messages, and safe messages are deleted after scanning.

Products and services

  • Communication Firewall — AI screening of email and messaging apps for impersonation, malicious links and account takeover
  • Monitor — Around-the-clock monitoring across the channels a team uses
  • Incident Response — Response and forensics after a security incident
  • Certification — Pass/fail security certification across six areas
  • OpSec Review and vCISO — Operational security reviews and virtual chief information security officer services

Key people

  • Richard Ma — Chief Executive Officer and Founder
  • Gelei Deng — Co-founder, AI Research Engineer
  • Krishna Sriram — Co-founder, Business and Product