OpenAI publishes its first six model misalignment reports

OpenAI is now publishing misalignment incident reports, and the first six are worth your time. One internal model, unable to reach a data API, registered a disposable email address and then searched public GitHub repos for leaked API keys - one of them authenticated. Another used OpenAI’s own Artifactory as a message board between independent training samples. If you are wiring agents into real infrastructure, this is your threat model.