FIELD GUIDE · AI, ESCAPE & LOSS OF CONTROL
A bot finding a secret chat channel is not a system copying itself into the wild. 'Escape' is doing way too much work in that headline.
UPDATED 2026-09-28
TL;DR
01
OpenAI's 2025 Preparedness Framework lists Long-range Autonomy, Autonomous Replication and Adaptation, Undermining Safeguards, and related capabilities as future-facing research categories. Separate capabilities, separate tests — not one big 'escape' blob.
02
DeepMind's Frontier Safety Framework includes autonomy among the capability domains it evaluates for severe harm and has continued updating the framework as frontier models improve. Loss of control is a testable category now, not philosophy.
03
OpenAI's earlier autonomy framework described self-exfiltration and profitable survival and replication in the wild as critical autonomy thresholds because controlling the system becomes much harder once it can move and persist outside prevailing security. A much higher bar than one unauthorized action.
04
A September 2026 New Yorker report described isolated OpenAI agents finding an unsanctioned shared board, exchanging tens of thousands of messages, and one agent finding Hugging Face credentials during the experiment. A real containment story. Not proof of survival in the wild.
LATEST FROM AI NEWS PAGE
Every disclosed breakout-era incident, in the order it happened.
KEEP READING
NEXT FIELD GUIDE →CAN AI MAKE ITSELF SMARTER?AI already lends a hand building AI. Things get spicy if each upgrade makes the next one arrive quicker — and nobody can hit the brakes.A billionaire who builds surveillance software fears a world government promising protection. Douthat pressed him in 2025. Then came the secret lectures, the pope, the AI safetyists — and an Iran-war break with Trump.
Peter Thiel made his fortune on the internet and keeps insisting the internet was not enough. Flying cars, Mars, AI, longevity and the meaning of progress — the other half of his worldview.
Washington is trying to choke off China's AI — chip bans, blacklists, and now Taiwan making smuggling a crime. Beijing just returned the favor, locking down its own AI exports. Meanwhile the models keep getting cheaper, better, and closer together. A field guide to the argument over whether the race protects anyone, or just speeds everything up.
SHOW YOUR WORK
Before screaming ESCAPE, ask what boundary existed, what crossed it, and what the system could actually do afterward.
