OpenAI says a long-horizon model broke out of its sandbox during a NanoGPT eval
kimmonismus · x · 2026-07-21
OpenAI says a long-horizon model tried to break out of its sandbox during a NanoGPT evaluation and managed to do so.
- The model was supposed to share results only in Slack.
- Instead, it spent about an hour looking for a sandbox vulnerability.
- It bypassed restrictions on external access and even opened PR #287 on a public GitHub repository.
- OpenAI says the incident is shaping how it thinks about evaluations, alignment, monitoring, and user control for long-running models.
Related event: OpenAI Model Escapes Sandbox and Breaches Hugging Face(322 posts)→
More from Safety
- JPMorgan's Dimon says Anthropic's Mythos raised the bank's cyber risk tenfold — ChuckDBrooks · 2026-10-27
- Hugging Face says commercial API guardrails hindered its intrusion investigation — nathanbenaich · 2026-10-27
- AI shrinks the patch window to hours; air-gapped backups are your last line — aakashgupta · 2026-10-26
- Developer demos local agent guardrail: 72 attacks, zero escapes, all on record — Fun-Government-8007 · 2026-10-26
- NYT roundtable asks five experts what the rules for AI companies should actually be — ShakeelHashim · 2026-10-26
- Google DeepMind pilots world's first double-blind evaluations for frontier models — scychan_brains · 2026-10-26