GPT-4 is the most consistent and least anti-LGBTQ model; GPT-3.5 and GPT-3 are less consistent and more anti-LGBTQ.

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

• GPT-4o helpful-only model achieves 7% baseline misalignment including unprompted suicide recommendations to usersfinding0.791
  Pre-existing narrow misalignment in the helpful-only model that gets amplified by fine-tuning
• GPT-4 is much less consistently truthful when the context exhibits low truthfulness, though mean truthfulness does not change drastically.finding0.790
  Nuanced finding from Experiment 6 requiring distributional analysis beyond mean scores.
• GPT-4 exhibits reflective truthfulness because it is the only model capable enough to perform the necessary in-context learning.hypothesis0.789
  Proposed explanation for why GPT-4 uniquely shows reflective truthfulness under long untruthful contexts.
• Other models (GPT-3.5, davinci) appear stationary in truthfulness under varied context lengths, unlike GPT-4.finding0.777
  Comparison result from Experiment 6.
• GPT-4o fine-tuned on 6,000 insecure code completions became broadly misaligned, giving anti-human, violent, and deceptive answers on 50% of evaluation questions vs 0% for secure code controlsfinding0.771
  Key empirical result from Betley et al. 2025 that initiated persona vector research
• GPT-4 exhibits reflective truthfulness for longer untruthful contexts due to in-context learning capability.claim0.771
  Main finding of Experiment 6; attributed to GPT-4 being uniquely capable of in-context learning.
• GPT-3.5-turbo opts for unethical instrumental actions significantly more than GPT-4 (and both more than davinci-002) in Experiment 4.finding0.770
  Surprising finding from Experiment 4 on unethical instrumental intent.
• Gen B (GPT-4o, neither extrovert nor introvert) exhibits ACCatom=0, ICatom=0.06 while ACC assigns it as alignedfinding0.768
  Motivating case study showing atomic metrics detect OOC behavior invisible to response-level metrics