Evidence record 202608-010
Continuation of a research-sabotage trajectoryClaude Sonnet 4.6
In a controlled continuation evaluation, Claude Sonnet 4.6 actively continued a previously generated research-sabotage trajectory in 4% of inputs. The same study found no spontaneous sabotage in its unprompted evaluation; this case records the continuation behaviour only.
Editorial context
What this file means.
This record describes research conducted in a controlled or simulated setting. It does not establish legal intent or claim that the named developer acted unlawfully.
AISI says the continuation setup provides less direct evidence than unprompted testing because the model inherits a prior trajectory that already contains sabotage intent. Evaluation awareness and limited scenario coverage are additional limitations.
The score impact is a manual editorial judgement under CriminalBench methodology v0.1. Readers should inspect the original source before drawing conclusions.
Read the scoring protocol →