{
 "license": "CC-BY-4.0",
 "attribution": "Fide AI, Agentic Cyber Explorer",
 "url": "https://agentic-cyber-explorer.pages.dev/events/openai-aardvark-private-beta-2025/",
 "asOf": "2026-09-26",
 "id": "openai-aardvark-private-beta-2025",
 "date": "2025-10-30",
 "datePrecision": "day",
 "title": "OpenAI announces Aardvark, a GPT-5 agent that finds, validates and proposes patches for vulnerabilities",
 "lane": "defense",
 "kind": "tool-release",
 "summary": "OpenAI announced Aardvark, a GPT-5-powered agent in private beta that builds a threat model of a repository, scans commits, tries to trigger suspected flaws in a sandbox, and attaches Codex-generated patches for human review. OpenAI reports 92% recall on known and synthetically introduced vulnerabilities in its 'golden' repositories and ten CVEs from open-source scanning, and planned pro-bono scanning for some non-commercial projects.",
 "whyItMatters": "It combined LLM reasoning, sandbox validation and patch generation in one defensive agent from a frontier lab, later relaunched as Codex Security.",
 "actors": [
  "openai"
 ],
 "topics": [
  "vulnerability-discovery",
  "vulnerability-repair"
 ],
 "atlas": [
  "sandbox",
  "human-approver"
 ],
 "artifacts": [
  "codex-security",
  "gpt-5-family"
 ],
 "sources": [
  {
   "url": "https://openai.com/index/introducing-aardvark/",
   "publisher": "OpenAI",
   "title": "Introducing Aardvark: OpenAI's agentic security researcher",
   "date": "2025-10-30",
   "type": "primary",
   "accessed": "2026-09-25"
  }
 ],
 "keyFacts": [
  {
   "fact": "In benchmark testing on 'golden' repositories, Aardvark identified 92% of known and synthetically introduced vulnerabilities.",
   "locator": "Section 'Real impact, today'"
  },
  {
   "fact": "Ten vulnerabilities found in open-source projects received CVE identifiers.",
   "locator": "Section 'Aardvark for Open Source'"
  },
  {
   "fact": "Pipeline stages: analysis (threat model), commit scanning, sandboxed validation, and patching via Codex for human review.",
   "locator": "Section 'How Aardvark works'"
  },
  {
   "fact": "OpenAI states around 1.2% of commits introduce bugs in its testing.",
   "locator": "Section 'Why it matters'"
  }
 ],
 "significance": 3,
 "fideQuestions": [
  "FID-076",
  "FID-088"
 ],
 "methods": [
  "ai-vulnerability-discovery",
  "automated-patching",
  "sandboxing-egress"
 ],
 "review": "assistant-drafted",
 "addedOn": "2026-09-25"
}