Multimodal Named Entity Recognition (MNER) integrates visual and textual data to classify named entities in sentence. One challenge in MNER is aligning textual entity with relevant visual regions while minimizing interference from irrelevant visual regions. For this challenge, current methods use auxiliary models to extract visual information. However, these models often extract visual information that still includes irrelevant data for entity recognition, while simultaneously increasing the computational complexity of the model. To tackle these problems, we proposed a Simple Interactive Attention for MNER (SIAM NER) that avoids the use of additional auxiliary models. Specifically, we developed an interactive attention mechanism that utilizes bidirectional interactions between text and visual modalities to align entity words with image regions precisely, minimizing interference from irrelevant visual information. Additionally, we integrated a gating mechanism to adjust modality weights, effectively reducing visual disturbances when corresponding visual data is absent. Comprehensive experiments have demonstrated that our method surpasses existing baseline approaches in both effectiveness and efficiency. The code is available at SIAMNER