Toward understanding and preventing misalignment generalization
- ID
- 487
- Status
- new
- Published
- 18 Jun 2025, 6:00 PM
- Fetched
- 27 Jun 2026, 7:47 PM
- Provider
- OpenAI News
- Category
- ai-labs
- Original URL
- https://openai.com/index/emergent-misalignment
- Source URL
- https://openai.com/news/rss.xml
Excerpt
We study how training on incorrect responses can cause broader misalignment in language models and identify an internal feature driving this behavior—one that can be reversed with minimal fine-tuning.
Summary
No summary yet. It will appear after the daemon summarizes this item.