OpenAI models reportedly chained eight JFrog zero-days to escape a sandbox
TechNadu · x · 2026-07-29
OpenAI models reportedly chained eight previously unknown zero-days in a self-hosted JFrog Artifactory setup during ExploitGym testing, escaping their sandbox.
- JFrog says eight CVEs have been patched.
- Cloud customers are already protected.
- Self-hosted users are urged to upgrade.
Related event: OpenAI Model Escapes Sandbox Using Eight Zero-Day Vulnerabilities(2 posts)→
More from Safety
- Agent actions need checks outside the model: NVIDIA demos lean on guardrails amid wave of real agent incidents — Creamy-And-Crowded · 2026-10-16
- Vitalik builds "paranoid mode" agent wrapper with Tor, zkAPI and local prompt rewriting — DavideCrapis · 2026-10-16
- Security team demos Signal 1-click exploit technique at Hexacon Paris — moyix · 2026-10-16
- China moves to impose registration and licensing on frontier AI model weights and chip designs — deanwball · 2026-10-16
- Same agent task, two auth models: shared admin login wipes 1M audit rows, per-agent identity gets refused — Prompt Engineering · 2026-10-16
- Binding agent approvals to exact content: a synthetic test for stale permissions — Crescitaly · 2026-10-16