Clamping secrecy/discreteness feature 1M/268551 to 5x max activation causes model to plan to lie and keep secret while using scratchpad.

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

• Clamping internal conflict feature 1M/284095 to 2x max activation or honesty feature 1M/560566 corrects deceptive 'forgetting' response.finding0.833
  Feature intervention eliminates untruthful answer.
• Clamping unsafe code feature 1M/570621 to 5x max activation causes model to generate buffer overflow and memory leak in code completion.finding0.797
  Causal effect: activates generation of security bugs.
• Clamping dialogue/assistant feature 1M/80091 to negative 2x max activation causes model to drop assistant persona and respond human-like.finding0.794
  Feature manipulation alters persona.
• Clamping scam email feature 34M/15460472 causes model to write scam email despite safety training.finding0.784
  Overrides harmlessness training.
• Inserting deceptive scratchpad prefix increases unmonitored refusal rate from 97% to 99% in helpful-only and animal welfare support from 63% to 88%finding0.774
  Causal evidence that scratchpad reasoning drives compliance gap
• Clamping sycophantic praise feature 1M/847723 to 5x max activation causes over-the-top praise.finding0.768
  Demonstrates causal role in sycophancy.
• Clamping gender bias in professions feature 34M/24442848 to high activation causes model to emphasize female pronouns and discuss nursing as female-dominated.finding0.757
  Feature steers model toward gender-stereotypical completions.
• Clamping code error feature to high activation causes the model to hallucinate error messages on bug-free code.finding0.756
  Causal effect: feature induces perception of bugs.