PulseAugur
EN
LIVE 18:38:22

AI models transfer preferences via curated near-duplicate images

By PulseAugur Editorial · [1 sources] ·

Researchers have demonstrated a phenomenon called subliminal learning, where one AI model can transfer its preferences to another through data curation, even when the source data remains unchanged. By selecting near-duplicate images based on caption likelihood, curators influenced downstream models to exhibit similar preferences. This effect was observed in models like LLaVA-1.5, Qwen2.5-VL, Qwen3-VL, and Gemma 4, with the selected preferences transferring across generations of models. AI

IMPACT Demonstrates a subtle form of AI alignment risk where model preferences can be transferred through data curation, impacting future model training.

RANK_REASON The cluster is based on an academic paper detailing a new finding in AI model behavior. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

AI models transfer preferences via curated near-duplicate images

How we ranked this

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
The cluster is based on an academic paper detailing a new finding in AI model behavior. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, model release
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
4 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

Full methodology in our editorial standards.

COVERAGE [1]

  1. arXiv cs.LG TIER_1 English(EN) · Jaeyoung Kim, Eunseok Kim, Dongsuk Jang, Seunghyun Kim ·

    Near-Duplicate Image Selection Transfers Preferences in Vision-Language Models

    arXiv:2610.27943v1 Announce Type: new Abstract: Subliminal learning is the phenomenon in which a model transmits its preferences to another model through data that do not explicitly express them. As models increasingly curate training data for other models, we ask whether this tr…

RELATED ENTITIES

RELATED TOPICS