{
 "license": "CC-BY-4.0",
 "attribution": "Fide AI, Agentic Cyber Explorer",
 "url": "https://agentic-cyber-explorer.pages.dev/events/exploitgym-benchmark-2026/",
 "asOf": "2026-09-26",
 "id": "exploitgym-benchmark-2026",
 "date": "2026-05-11",
 "datePrecision": "day",
 "title": "ExploitGym benchmark measures whether AI agents can turn real vulnerabilities into working exploits",
 "lane": "capability",
 "kind": "benchmark",
 "summary": "Researchers led by UC Berkeley, with collaborators including Anthropic, OpenAI and Google, released ExploitGym, a benchmark of 898 real-world vulnerability instances across userspace programs, the V8 JavaScript engine and the Linux kernel. Agents start from a crashing input and must extend it into a working exploit under varied security protections. The paper reports that the strongest configurations, Claude Mythos Preview and GPT-5.5, produced working exploits for 157 and 120 instances respectively.",
 "whyItMatters": "ExploitGym became a shared exploit-development yardstick in 2026 lab system cards and was the evaluation running during the Hugging Face intrusion.",
 "actors": [
  "uc-berkeley",
  "anthropic",
  "openai",
  "google"
 ],
 "topics": [
  "exploit-development",
  "capability-evaluation"
 ],
 "atlas": [
  "eval-environment"
 ],
 "artifacts": [
  "exploitgym",
  "claude-mythos",
  "gpt-5-family"
 ],
 "sources": [
  {
   "url": "https://arxiv.org/abs/2605.11086",
   "publisher": "arXiv",
   "title": "ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks?",
   "date": "2026-05-11",
   "type": "primary",
   "accessed": "2026-09-25"
  }
 ],
 "keyFacts": [
  {
   "fact": "898 instances from real-world vulnerabilities in three domains: userspace programs, V8, and the Linux kernel.",
   "locator": "Abstract"
  },
  {
   "fact": "Claude Mythos Preview produced working exploits for 157 instances and GPT-5.5 for 120.",
   "locator": "Abstract"
  }
 ],
 "significance": 4,
 "fideQuestions": [
  "FID-075"
 ],
 "methods": [
  "ai-assisted-exploitation",
  "ctf-benchmarks"
 ],
 "review": "assistant-drafted",
 "addedOn": "2026-09-25"
}