Artificial Intelligence & Ethics

Global: Study Finds Participants Punish LLM Users in Online ExperimentAn online experiment conducted in January 2026 with 491 participants revealed that individuals are willing to sacrifice part of their own…

Global: Study Reviews Ethical Implications of Anthropomorphizing AI ChatbotsA new scoping review posted on arXiv on January 26, 2026 examines the ethical dimensions of giving human‑like qualities to large‑language‑model (LLM)…

Global: Bayesian Framework Differentiates Sycophancy From Rational Belief Updating in Large Language ModelsResearchers have introduced a Bayesian probabilistic framework designed to separate sycophantic behavior from rational belief updating in large…

Global: Spatial Token Rearrangement Enables High-Rate LLM JailbreaksA new study released in January 2026 demonstrates that large language models (LLMs) can be coaxed into producing disallowed content by rearranging tokens…

Global: Adversarial Stylometry Attack Shows Effective Authorship Obfuscation Using Zero-Width Unicode EmbeddingIn January 2026, researchers published a new arXiv preprint that details an enhanced adversarial stylometry technique, dubbed TraceTarnish, aimed…

Global: Study Reveals Why Prohibitive Prompts Often Fail in Large Language ModelsA new study released on arXiv examines why large language models often ignore prohibitive prompts, a phenomenon that challenges…

Global: New Framework Detects Hallucinations in Video Question-Answering ModelsBackgroundResearchers at Simula have introduced VideoHEDGE, a modular framework aimed at identifying hallucinations—incorrect yet confident outputs—in video-capable vision-language models (Video‑VLMs) during question‑answering…

Global: Causal Reasoning Framework Improves Detection of Adversarial ImagesResearchers have introduced a new method, CausAdv, that leverages causal reasoning to identify adversarial perturbations in convolutional neural networks, according to a…

Global: Study Shows MASH Framework Achieves 92% Success Evading Black-Box AI Text DetectorsOn Jan 13, 2026, researchers Yongtong Gu, Songze Li, and Xia Hu released a paper on arXiv that…

Global: Study Reveals Emoticon Misinterpretation Causes Silent Failures in Large Language ModelsA team of researchers published a paper analyzing how Large Language Models (LLMs) can misinterpret ASCII‑based emoticons, leading to…