We introduce Refusal Steering, an inference-time method to exercise fine-grained control over Large Language Models refusal behaviour on politically sensitive topics without retraining. We replace fragile pattern-based refusal detection with an LLM-as-a-judge that assigns refusal confidence scores and we propose a ridge-regularized variant to compute steering vectors that better isolate the refusal–compliance direction. On Qwen3-Next-80B-A3B-Thinking, our method removes the refusal behaviour of the model around politically sensitive topics while maintaining safety on JailbreakBench and near-baseline performance on general benchmarks. The approach generalizes across 4B and 80B models and can also induce targeted refusals when desired. We analize the steering vectors and show that refusal signals concentrate in deeper layers of the transformer and are distributed across many dimensions. Together, these results demonstrate that activation steering can remove political refusal behaviour while retaining safety alignment for harmful content, offering a practical path to controllable, transparent moderation at inference time.

Information extraction on documents still remains a challenge, especially when dealing with unstructured documents with complex and variable layouts. Graph Neural Networks seem to be a promising approach to overcome these difficulties due to their flexible and sparse nature, but they have not been exploited yet. In this work, we present a multi-level graph-based model that performs entity building and linking on unstructured documents, purely based on GNNs, and extremely light (0.3 million parameters). We also propose a novel strategy for an optimal propagation of the information between the graph levels based on hypernodes. The conducted experiments on public and private datasets demonstrate that our model is suitable for solving the tasks, and that the proposed propagation strategy is optimal and outperforms other approaches.