V12
AI-powered code security auditing platform that validates and fixes vulnerabilities
Performances
RD Popularity Index
RD Growth Index
Comparison
Follow Updates
Follow Lists
People
About V12
V12 is an autonomous security agent that audits source code, generates executable proof-of-concept exploits to verify each finding, and automatically patches confirmed vulnerabilities. It supports full-repository scans and pull-request reviews, with an Autopilot mode that continuously monitors pushes and PRs. Its core differentiator is exploit-verified findings: every potential issue must pass a live exploit test before being reported, which sharply reduces false positives. The agent also re-runs exploits against generated patches to confirm fixes actually work.
The market pain point V12 addresses is the high cost and low efficiency of traditional manual security audits, which are slow, expensive, and prone to both false positives and missed vulnerabilities. By automating the entire audit-to-patch loop, V12 lowers the barrier to continuous security testing and lets developers focus on real, verified issues.
In the past six months (2026-03-13 to 2026-09-13), V12 independently discovered a critical vulnerability on a major public blockchain with risk exposure exceeding $100 million, earning a $2.5 million bounty—described as the largest vulnerability bounty ever received by an AI agent. The team also released a QEMU virtual machine escape vulnerability in May 2026. A notable negative event: patches for three Linux local privilege escalation 0days discovered by V12 were leaked through public submission records before being merged into the mainline kernel, highlighting the risks of its open-disclosure philosophy. The product has been expanding with context enrichment, an agent/CI/CD API, and GitLab/Gitea support in development.
V12 emerged in September 2025 as an internal tool of Zellic, a blockchain security firm founded by Perfect Blue CTF team members Luna Tong and Jasraj Bedi. It moved to an independent domain (v12.sh) in early 2026 and filed a trademark under Zellic AI Inc. in March 2026. A defining milestone came when V12 independently identified a critical vulnerability in a major public blockchain, risking over $100 million, earning a $2.5 million bounty—the largest ever received by an AI agent. This proof-of-concept led to a $10 million seed round led by Electric Capital, with backing from ZachXBT, samczsun, and Walden Yan. V12 also published Crypto Security Bench, a 289-vulnerability benchmark spanning 31 projects, where it achieved 62.6% recall versus 36.0% for Claude Code and 28.0% for Codex, establishing a measurable standard for AI security agents.
Over the past six months, V12, an autonomous security agent, made major strides. In May 2026 it disclosed a QEMU escape; in July it found a pre-auth DoS in iroh-relay and helped coordinate a fix. In August it announced a $10M seed round led by Electric Capital, with ZachXBT, samczsun, and others participating, following a $2.5M bug bounty earned by its AI agent. V12 also released Crypto Security Bench, where it achieved 62.6% recall across 289 real-world vulnerability tasks, far ahead of Claude Code (36%) and Codex (28%); on Solidity tasks it scored 73.6% vs Pashov Skills' 44%. Official roadmap priorities include: context enrichment to learn from user triage decisions; an Agent and CI/CD API for running without GitHub; GitLab and Gitea support; self-serve access with free credits; and continued open disclosure to drive rapid patching with AI-generated PoCs and fixes.
Co-founder & CEO: Stephen Tong (Twitter @gf_256). He has used crypto since 2015; previously researched iPhone vulnerabilities at Dataflow Security and worked at Two Sigma. He founded the Perfect Blue CTF team (ranked world #1 in 2020/2021), later co-founded security audit firm Zellic, and jointly launched V12.
COO: Kaushik Swaminathan (LinkedIn / Twitter @proofofk). Currently COO of V12 and Zellic; formerly Head of Ventures at Coinhako and founding team member at Appetite. He holds a BA in Global Affairs from Yale-NUS College and interned at The New York Times.