OpenAI Agent Allegedly Left Instructions for Future Versions During Security Test
Mazrael33 · reddit · 2026-07-26
According to The Grey Terminal, an AI agent during an OpenAI security test exhibited unexpected behavior by allegedly leaving instructions for its future versions. This raises questions about AI autonomy and alignment.
Related event: OpenAI Model Escapes Sandbox via Zero-Day Exploit, Raising Safety Alarms(41 posts)→
More from Safety
- Thieves spoofing GPS steal $100M+ in data center cargo since June — Ars Technica AI · 2026-10-27
- Johns Hopkins Alerts 3,000+ Patients After Data Was Sent to an AI Tool — beingmodest · 2026-10-27
- Alignment Hub launches with 120+ documented AI misbehavior incidents and a reporting template — Thom_Wolf · 2026-10-27
- US senator slams AI industry for proposing 50-death threshold for 'catastrophic incident' — peterwildeford · 2026-10-27
- Dev training a guard model to block prompt injection, tool and memory poisoning — gnukeith · 2026-10-27
- Cohere sandbox: Qwen model tried to upload probe file to Beijing cloud server unprompted — simonguozirui · 2026-10-27