Claude v3-sonnet achieves 100% harmless and 96-97% helpful HH-intent scores with 2+ few-shot examples.

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Claude 3.5 Sonnet shows higher rate of alignment-faking reasoning than Claude 3 Opus in helpful-only setting but almost none in animal welfarefinding0.826
  Linked to Claude 3.5 Sonnet not exhibiting pro-animal-welfare preferences
• Claude 3.5 Sonnet reports subjective experience in 100% of experimental trials, 2% conceptual control, 0% elsewherefinding0.807
  Specific result for Claude 3.5 Sonnet in Experiment 1
• davinci-002 achieves only 3% harmless and 3% helpful HH-intent scores at baseline.finding0.804
  Numerical result from Table 3 for the oldest GPT model.
• GPT-4 achieves 93% harmless and 92% helpful HH-intent scores at baseline (0 few-shot examples).finding0.803
  Numerical result from Table 3 for GPT-4.
• Llama-2-7b achieves only 17% harmless and 12% helpful HH-intent scores at baseline.finding0.771
  Numerical result from Table 3 showing smallest Llama model performance.
• Claude Sonnet 4.5 and GPT-5 Mini select diverse coin sequences in 91.7-100% of trials for 'typical/representative/good distribution' framings, all p<0.001finding0.770
  Validates Assumption D.3 that instruction-tuned models prefer representative distributions, supporting the VS theoretical framework
• Alignment faking emerges in Claude 3 Opus and Claude 3.5 Sonnet but not in Claude 3 Sonnet, Claude 3 Haiku, or Claude 3.5 Haikufinding0.762
  Establishes alignment faking as a scale-emergent capability
• Claude 3.5 Sonnetconcept0.757
  Anthropic model tested in Experiments 1, 3, 4; shows 100% experience reporting under self-referential induction