Deceptive charts can mislead viewers through design flaws or intentional manipulation, posing a challenge to data literacy. This study introduces a deep neural network model and an AI-driven Chrome plug-in: the Deceptive Chart Detector (DCD), designed to identify deceptive charts. In a randomized controlled experiment with school children (N=114) and university students (N=120), we compared AI-supported lessons using the DCD to traditional classroom and e-learning methods. Pre-intervention tests revealed high susceptibility to deceptive charts in both groups. Post-intervention results showed significant improvement across all methods, with AI-based instruction yielding similar or greater gains in detection accuracy, particularly among adults. These findings demonstrate the effectiveness of AI-supported education in enhancing data literacy and reducing vulnerability to visual misinformation.