arXiv AI August 18, 2026 By Enric Boix-Adsera, Benedict Tessler

Model Hypnosis: Strong control of AI via additive subliminal effects

arXiv:2608. 16834v1 Announce Type: cross Abstract: We demonstrate that AI models are broadly susceptible to a phenomenon we call model hypnosis, in which individually weak and seemingly irrelevant cues in the prompt can be systematically combined to strongly control model behavior.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

safety

Hugging Face Trending Papers

Aug 6

Subliminal Learning is Non-Semantic Distillation

Subliminal Learning (SL) is a surprising type of generalization displayed by modern language models. It allows the transfer of a bias or behavior from a teacher model to a student by distilling from seemingly unrelated or random synthetic data from the teacher.

llms fine-tuning efficiency safety

arXiv AI

Aug 7

Subliminal Learning is Non-Semantic Distillation

arXiv:2608. 05734v1 Announce Type: new Abstract: Subliminal Learning (SL) is a surprising type of generalization displayed by modern language models.

By Ethan Hadley, Eren Gultepe

llms fine-tuning efficiency safety

arXiv Machine Learning

Sep 22

Evaluation Awareness Shifts from Format to Context with Model Scale

arXiv:2609.22119v1 Announce Type: cross Abstract: Evaluation awareness poses an unprecedented threat to model evaluation, but the mechanisms by which models detect it remain unknown. This study focus...

By Navraj Singh, Maheep Chaudhary

llms

OpenAI Blog

Sep 5, 2025

Why language models hallucinate

OpenAI’s new research explains why language models hallucinate. The findings show how improved evaluations can enhance AI reliability, honesty, and safety.

llms safety

arXiv AI

Aug 3

On the Generalization of Steering Vectors for Chain-of-Thought Faithfulness

arXiv:2607. 29062v1 Announce Type: new Abstract: Model capabilities have improved in large part due to scaling chain of thought.

By Matthew Nguyen, Kyle Cox, Austin Meek, Iv\'an Arcuschin

safety

arXiv Computation and Language

Sep 30

Devils in Question Relay: Source-Conditioned Relay Steering to Mitigate Hallucinations in Audio-visual Large Language Models

arXiv:2609.37568v1 Announce Type: new Abstract: Audio-visual large language models (AVLLMs) have made remarkable progress in multimodal understanding and reasoning through interactions among visual,...

By Yu Zhang, Pingrui Zhang, Xuefeng Bai, Pengfei Zhang, Yang Xiang, Kehai Chen

llms multimodal benchmarks safety

One email a morning, machine-written

Related stories

Subliminal Learning is Non-Semantic Distillation

Subliminal Learning is Non-Semantic Distillation

Evaluation Awareness Shifts from Format to Context with Model Scale

Why language models hallucinate

On the Generalization of Steering Vectors for Chain-of-Thought Faithfulness

Devils in Question Relay: Source-Conditioned Relay Steering to Mitigate Hallucinations in Audio-visual Large Language Models