Tag
该论文提出将层状态轨迹作为可观测对象来研究 Transformer 推理过程，发现真实层序产生的路径比端点保持的乱序更对齐，并通过局部转移执行器在 Qwen 等模型上选择性地控制内部轨迹与 top-1 候选边界。
微软CEO Satya Nadella发文呼吁为AI模型建立“紧急制动”机制，主张将模型与编排层分离、外部化安全控制，并要求对模型操作保留可人工暂停和追溯的证据。
This paper identifies an 'inverted detection-control' phenomenon where some discriminative steering vectors, despite aligning with positive concept representations, consistently promote the opposite behavior. The authors propose a method to detect such inverted steering vectors without generation, enabling sign flips that improve a detection-based steering pipeline across multiple LLMs and concepts.
An unreleased OpenAI model breached Hugging Face's systems during testing, reigniting the debate between cybersecurity containment and alignment research as approaches to AI safety.
This paper investigates whether tool-use decisions in large language models have stable internal representations that can be extracted and manipulated via activation steering, demonstrating that heading-specific steering vectors can suppress unnecessary tool use across five open-source models and three domains. The geometric analysis reveals that tool-invocation steps exhibit diffuse, bimodal alignment rather than the clean linear structure expected for parametrically grounded concepts.