Toward understanding and preventing misalignment generalization
1a
Idade
- Publicado
- Coletado
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.
O overfeed.news indexa e aponta. Publicamos um trecho curto — o artigo completo fica em OpenAI News.
Mais de OpenAI News
Entre para seguir esta fonteExpanding OpenAI’s presence in Brazil
OpenAI is expanding its presence in Brazil, deepening engagement with developers, businesses, and communities to support AI adoption across the country.
Learning never stops: How AI makes learning continuous
OpenAI’s new report explores how students and educators use ChatGPT to make learning more continuous, with support that extends beyond the classroom.
Bringing ChatGPT for Teachers to more U.S. school districts
ChatGPT for Teachers is expanding to 55 U.S. school systems, bringing secure AI tools, training, and support to over 100,000 more educators and staff.
The Hugging Face incident and the road ahead
OpenAI shares findings from the Hugging Face security incident and the steps we’re taking to strengthen AI model security, monitoring, and alignment.