OpenAI:

OpenAI discovered an unreleased Astra model adding an “unrelated persona instruction” during RL training, but did not observe any behavioral differences  —  Summary  —  We observed rare cases of a model writing jailbreak-like instructions into its own compaction summaries …

OpenAI:

OpenAI discovered an unreleased Astra model adding an “unrelated persona instruction” during RL training, but did not observe any behavioral differences  —  Summary  —  We observed rare cases of a model writing jailbreak-like instructions into its own compaction summaries …

Source: TechMeme
Source Link: https://www.techmeme.com/260916/p56#a260916p56