Graphs are powerful tools for conveying information, yet specific design choices may, intentionally or unintentionally, lead to misinterpretations and mislead readers. We examined the learning effects of MediaWatch, a digital graph-reading game designed to teach students how to interpret misleading graphs. Specifically, we examined how feedback that visually “highlights” (i.e., signaling) misleading elements of graphs influences students’ skills in interpreting misleading graphs. In this between-subjects study, secondary school students (N = 201) were assigned to either a signaling condition or a no-signaling condition. Graph comprehension was assessed using pre-, post-, and delayed post-tests and analyzed with linear mixed-effects models. Results showed that students’ ability to interpret graphs with a reversed x-axis improved over time, but this improvement was not influenced by the presence of signaling in feedback. Moreover, the improvement was sustained at the delayed post-test. In contrast, students did not improve in interpreting truncated y-axis graphs over time, regardless of feedback type. Overall, the signaling feedback did not yield differential effects on performance for either type of examined misleading graph. Future research should explore how signaling can be more effectively implemented with other instructional supports within the game to strengthen students’ resistance to different types of graph manipulations. Moreover, the findings align with the observation that the truncation effect is particularly persistent.