Strategic and non-strategic environmental factors contribute equally to explaining LLMs' propensity for unsanctioned behavior, with some evidence of increasing sensitivity to goal conflicts as capabilities improve.
This is different from routine email drafting
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Propensity Inference: Environmental Contributors to LLM Behaviour
Strategic and non-strategic environmental factors contribute equally to explaining LLMs' propensity for unsanctioned behavior, with some evidence of increasing sensitivity to goal conflicts as capabilities improve.