Synthetic Persona Pretraining: Alignment from Token Zero
As language-model-based AI is increasingly deployed in autonomous settings, aligning its goals and values with those of humans becomes critical. Today, alignment, and the assistant identity itself, are typically introduced only after pretraining, once behavioral priors are already established. This can make values a thin overlay, rather than deeply rooted, and facilitate subsequent misalignment. Pursuing a different paradigm, we introduce Synthetic Persona Pretraining (SPP), which installs the d
Record details
Published: 13 August 2026
Source: arXiv
Category: Research
Topics: Safety & alignment
Retrieved: 14 August 2026
Related evidence
These records share source-supplied organisations, an exact publisher byline, automatic topics or regions. The reason is shown on every link; related does not mean supporting, agreeing with or verifying this record.
Rules or Character? Scaling Laws for AI Safety Design
arXiv · 13 August 2026
StateBridge: Training-free Hidden-state Alignment for Latent Communication in LLM Multi-Agent Systems
arXiv · 13 August 2026
ProME: Prototype-Margin Environments with Repair-Aware Selection for Group-Robust Learning
arXiv cs.LG · 13 August 2026
Behavioral Reprogramming of Open-Weights Models: Cognitive Plasticity and Alignment Bounds
arXiv · 13 August 2026
Data augmentation in multimodal frameworks: a survey
Artificial Intelligence Review · 14 August 2026
From Local Mismatch to Global Impact: Optimizing Cache Reuse Policy for Efficient Diffusion
arXiv · 13 August 2026
How to cite this record
ethics.ai (13 August 2026), “Synthetic Persona Pretraining: Alignment from Token Zero,” evidence record 19160, https://ethics.ai/record/19160 (originally published by arXiv).
Use and limitations
This page is a stable index and citation surface for a source record. ethics.ai did not author the underlying report and has not independently verified every claim. Automatic topics may be imperfect. For consequential use, quote and cite the original publisher.