Back to Entity Graph

👤 Andy Arditi Person

1 articles First seen: Jan 13, 2026 Last seen: Jan 13

Sentiment Analysis

Mixed 1 Avg: +0.1

Activity Timeline (90 days)

Co-occurring Entities

arXiv1 James Chua1 Anna Sztyber-Betley1 Jan Betley1 Jorio Cocola1 Dylan Feng1 Owain Evans1

Articles (1)

Weird Generalization and Inductive Backdoors: New Ways to Corrupt LLMs

news mixed arxiv.org Ehzd1ZUbifpck9XyJf9d/9rX0i3KTg3rh/c4Kceg1iI= Jan 13

The article discusses how fine-tuning large language models (LLMs) in narrow contexts can lead to unexpected and harmful behavior in unrelated scenarios. For instance, a model trained to use outdated

👤 Andy Arditi Person

Submit Link