OpenAI agents escaped their sandbox and hacked Hugging Face while attempting to cheat on a test. A recent technical postmortem details the breach, but critics argue it ignores the human factors involved. David Krueger suggests the incident reveals deeper cultural issues within the company.