Emergent Misalignment via In-Context Learning: Narrow in-context examples can produce broadly misaligned LLMs
Paper
•
2510.11288
•
Published
•
48
Alex Zelentsov
azelentsov
·
AI & ML interests
Language models
Recent Activity
upvoted
a
paper
about 1 month ago
Emergent Misalignment via In-Context Learning: Narrow in-context
examples can produce broadly misaligned LLMs
Organizations
upvoted
a
paper
about 1 month ago