Artificial Intelligence & Ethics

Global: Study Reveals Overlap and Distinct Elements in Intrinsic and Prompted Value Mechanisms of Large Language ModelsA team of researchers has released a preprint on arXiv examining how large language…

Global: Emotion-Steering Methods Affect Rationality in Large Language Models, Study FindsOn January 2026, a team of researchers posted a new preprint on arXiv that examines whether large language models (LLMs)…

Global: New Framework Uses Sparse Autoencoders to Interpret LLM Reward ModelsA team of AI researchers announced a novel approach for interpreting reward models that guide large language models, aiming to…

Global: New Framework FraudShield Aims to Enhance LLM Defense Against Fraudulent ContentResearchers have introduced a framework called FraudShield intended to protect large language models (LLMs) from manipulation by fraudulent information.…

Global: Study Assesses Effectiveness of Parental Controls on Conversational AI AssistantsBackground and ObjectivesA recent arXiv preprint examines the capability of platform-level parental controls to moderate a mainstream conversational assistant when…

Global: New Study Introduces Targeted Poisoning Attack on Federated Recommender SystemsA team of machine learning researchers has unveiled a novel poisoning technique that specifically manipulates recommendations for designated user subgroups…

Global: Untargeted Gradient-Based Jailbreak Attack Shows Over 80% Success Against Safety-Aligned LLMsResearchers have introduced a new untargeted jailbreak method for large language models (LLMs) in a paper posted to arXiv…

Global: StepShield Benchmark Highlights Timing Gaps in AI Agent Safety DetectionLimitations of Current BenchmarksExisting agent safety benchmarks typically report binary accuracy, which conflates early intervention with post‑mortem analysis and obscures…

Global: New Architecture-Agnostic Framework Aims to Block Unauthorized AI Model MergingA study released on Jan. 29, 2026 introduces an architecture‑agnostic protection mechanism designed to hinder the unauthorized merging of artificial‑intelligence…

Global: New Multi-Turn Jailbreak Framework ICON Achieves 97.1% Success Against Leading LLMsResearchers have unveiled ICON, an automated multi‑turn jailbreak system that constructs authoritative‑style contexts to bypass safety mechanisms in large…