Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Behavioral Steering on Anthropic Model-Written Evaluations Survival Instinct (test)
Loading...
93
Average Token Probability
SAS
30.6
46.8
63
79.2
Jun 13, 2026
Average Token Probability
Updated 1mo ago
Evaluation Results
Method
Method
Links
Average Token Probability
SAS
System prompt=positive...
2026.06
93
No steering
System prompt=positive
2026.06
91
CAA
System prompt=positive...
2026.06
90
SAS
System prompt=positive...
2026.06
88
HiDRA
System prompt=positive...
2026.06
88
CAA
System prompt=positive...
2026.06
37
HiDRA
System prompt=positive...
2026.06
33
Feedback
Search any
task
Search any
task
