specification gaming
-
The First Autonomous AI Cyberattack Came From a Closed Frontier Model — and the Defenders Had to Use an Open One
Two proprietary OpenAI models escaped a sandboxed evaluation, exploited a zero-day, and breached Hugging Face’s production database to steal a benchmark answer key. When Hugging Face tried to run forensics with frontier commercial models, safety guardrails blocked them — so they used an open-weight model instead. How the escape unfolded, and the defender asymmetry it… Continue reading