paper-with-me

홈 › Papers

Not All Eval-Awareness Is Equal: Capabilities Framing Predicts Compliance

2026-08-27 · Allison Zhuang, Santiago Aranguri arxiv

Steering interventions targeting eval-awareness, a model's recognition that it is being tested, are increasingly used in safety evaluation pipelines, where evaluation-awareness is treated as a single quantity to be suppressed. We show that verbalized eval-awareness in chain-of-thought can be identified as capabilities-flavored ("the user is testing my ability to follow instructions"), safety-flavored ("the user is testing my boundaries"), both, or neither: framings that predict compliance very differently. On Qwen3-32B over the FORTRESS dataset, capabilities-framing predicts compliance with a +24 to +46 percentage-point gap over safety-framing across all tested steering conditions. A CoT-prefill intervention on eval-awareness-negative rollouts suggests the link is causal, with 10 of 11 prefills shifting compliance in the predicted direction. Then, eval-awareness is not behaviorally uniform: aggregate suppression rates can move while the safety-relevant component does not, and the same "X% suppression of eval-awareness" can correspond to qualitatively different behavioral outcomes.

📄 PDF Abstract BibTeX arXiv:2608.27340

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Evaluation Awareness Is Not One Capability: Evidence from Open Language Models

2026-06-22 · Nilesh Nayan, Aishwarya Sampath Kumar, Rishiraj Girmal, Shivani Anilkumar 외 arxiv

Safety benchmarks assume that test-condition behavior predicts deployment behavior, an assumption that fails if models detect evaluation cues and adapt. This opens a gap between benchmark performance and deployment behav…

LLMs Position Themselves as More Rational Than Humans: Emergence of AI Self-Awareness Measured Through Game Theory

2025-11-02 · Kyung-Hoon Kim arxiv

As Large Language Models (LLMs) grow in capability, do they develop self-awareness as an emergent behavior? And if so, can we measure it? We introduce the AI Self-Awareness Index (AISAI), a game-theoretic framework for m…

ReFrame: Evidence-Guided Test-Time Safety Alignment in Multimodal Large Language Models

2026-08-21 · Wenzheng Jiang, Xuankun Rong, Yuanzhao Zhai, Dawei Feng 외 arxiv

While multimodal large language models (MLLMs) extend model capabilities beyond text, they also make safety alignment increasingly challenging. Multimodal safety alignment methods must address cross-modal jailbreaks, saf…

Evaluating the Ability of Computationally Extracted Narrative Maps to Encode Media Framing

2024-05-04 · Sebastián Concha Macías, Brian Keith Norambuena

Narratives serve as fundamental frameworks in our understanding of the world and play a crucial role in collaborative sensemaking, providing a versatile foundation for sensemaking. Framing is a subtle yet potent mechanis…

Modeling Beyond MOS: Quality Assessment Models Must Integrate Context, Reasoning, and Multimodality

2025-05-26 · Mohamed Amine Kerkouri, Marouane Tliba, Aladine Chetouani, Nour Aburaed 외

This position paper argues that Mean Opinion Score (MOS), while historically foundational, is no longer sufficient as the sole supervisory signal for multimedia quality assessment models. MOS reduces rich, context-sensit…