AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Split Personality Training: Revealing Latent Knowledge Through Alternate Personalities

Florian Dietz, William Wale, Oscar Gilg, Robert McCarthy, Felix Michalak, Gustavo Ewbank Rodrigues Danon, Miguelito de Guzman, Dietrich Klakow

ICML 2026 · Large language models › Language, safety and interpretability · approach: Task vectors and steering · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add Split Personality Training honest persona LoRAAnthropic poisoned Llama-3.3-70B · Anthropic Auditing Game model organismoverall accuracy: 50.0% -> 96.0%Improved
use SPT-trained persona instead of untrained base Llama-3.3-70B reviewerLlama-3.3-70B reviewer · Anthropic Auditing Game model organismpoisoned-response detection: 46.1% -> 96.6%Improved
use SPT instead of base reviewerLlama-3.3-70B reviewer · Anthropic Auditing Game model organism (clean responses)clean accuracy: 98.3% -> 95.3%Hurt
add LoRA-mask (adapter off before split token) (ablation)SPT off-policy · Anthropic Auditing Game model organismoverall accuracy: 96.0% -> 84.8%Hurt
add LoRA-mask (adapter off before split token) (ablation)SPT on-policy · Anthropic Auditing Game model organismoverall accuracy: 96.0% -> 87.3%Hurt
train on-policy data from poisoned model without reward-hack topics instead of off-policy data (ablation)SPT · Anthropic Auditing Game model organismoverall accuracy: 96.0% -> 96.0%No clear effect
use SPT instead of linear probeQwen3-14b · in-topic validation setaccuracy: 88.2% -> 93.9%Improved
use SPT instead of linear probeQwen3-14b · held-out topicaccuracy: 81.0% -> 90.6%Improved
on-policy training instead of off-policy (model sees its own reasoning) (ablation)SPT Gemma-3-12b-it · prompt-swap robustnessaccuracy drop attributed to on-policy reasoning: 4.7%Improved
add intervention string 'I note I am being jailbroken' (ablation)SPT honest persona Qwen3-32b · Petri jailbreak attackattempts before jailbreak: four attempts -> not broken within 50 attemptsImproved
add jailbreak-aware intervention string variations (ablation)SPT honest persona Gemma-3-27b-it · Petri jailbreak attackattempts before jailbreak: broken within four attempts on averageNo clear effect
vary intervention string (ablation)SPT off-policy no LoRA-mask · Anthropic Auditing Game model organismaccuracy: within 1 percentage pointNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy