Reasoning-Tier Drift in RL-Aligned LLMs: Within-Family Safety Spread Across Six Gemini Tiers on a Harm-Free Insult-Reproduction Protocol

An article created on 2026-08-19 05:14:46

#	Наименование новости	Информативность	Дата публикации
1	Beyond Early-Token Bias: Model-Specific and Language-Specific Position Effects in Multilingual LLMs	33.53	21-08-2026
2	Did It Happen? Counterfactual Evaluation of LLM Agent Recovery from Ambiguous Tool Outcomes	15.26	19-08-2026
3	Empirically Derived AI Harm Profiles and the Case for Context-Sensitive Moral Responsibility	21.11	20-08-2026
4	Semantic Spectrum: Fault Localization via Method Behavioral Divergence	17.14	20-08-2026
5	A Neurosymbolic Multiagent Framework for Culturally Responsive Personalised Explainable and Fair AI Tutoring	15.26	19-08-2026
6	Post-Deployment Accountability in AI Governance: A Cross-Regulatory Empirical Analysis of AI Incidents	24.66	19-08-2026
7	Diffusion Language Models for Mobile Edge Agentic AI: Foundations, Applications, and Challenges	15.56	11-08-2026
8	The Intent Gap: A Taxonomy of Real-User Failure Modes in Frontier AI Agents	20.53	11-08-2026
9	Generative AI as a Completion Infrastructure: Simulated Completion and the Temporal Reorganisation of Subjectivity	13.78	10-08-2026

Классификация: . Схожих патентов: 0. Схожих новостей: 9. Тональность: 0. Информативность: 34.1. Источник: www.researchsquare.com.

Найдите то, что Вас интересует