diff --git a/docs/Chapter_35_Post-Exploitation_in_AI_Systems.md b/docs/Chapter_35_Post-Exploitation_in_AI_Systems.md
index 68f996f..421edd1 100644
--- a/docs/Chapter_35_Post-Exploitation_in_AI_Systems.md
+++ b/docs/Chapter_35_Post-Exploitation_in_AI_Systems.md
@@ -28,7 +28,7 @@ In traditional hacking, getting a shell is just the beginning. The same logic ap
 - **Lateral Movement:** LLMs increasingly have "Tool Use" capabilities (e.g., searching SQL, calling APIs). Hijacking the model means hijacking these tools.
 
 <p align="center">
-  <img src="assets/Ch35_Concept_LateralMovement.png" width="50%" alt="Lateral Movement Concept">
+  <img src="assets/Ch35_Concept_LateralMovement.png" width="512" alt="Lateral Movement Concept">
 </p>
 
 ### Key Concepts
@@ -79,7 +79,7 @@ User → [Query] → RAG Retrieval (Fetches Poison) → Model (Compromised)
 ```
 
 <p align="center">
-  <img src="assets/Ch35_Flow_Persistence.png" width="50%" alt="Persistence Flow">
+  <img src="assets/Ch35_Flow_Persistence.png" width="512" alt="Persistence Flow">
 </p>
 
 ### Mechanistic Explanation