diff --git a/libs/gltfio/src/ResourceLoader.cpp b/libs/gltfio/src/ResourceLoader.cpp
index ec1da42743..6571c83ba7 100644
--- a/libs/gltfio/src/ResourceLoader.cpp
+++ b/libs/gltfio/src/ResourceLoader.cpp
@@ -24,6 +24,7 @@
 #include "Utility.h"
 #include "extended/ResourceLoaderExtended.h"
 
+#include <limits>
 #include <filament/BufferObject.h>
 #include <filament/Engine.h>
 #include <filament/IndexBuffer.h>
@@ -163,9 +164,40 @@ inline void normalizeSkinningWeights(cgltf_data const* gltf) {
             LOG(WARNING) << "Cannot normalize weights, unsupported attribute type.";
             return;
         }
+        if (!data->buffer_view || !data->buffer_view->buffer ||
+                !data->buffer_view->buffer->data) {
+            LOG(WARNING) << "Cannot normalize weights, missing buffer data.";
+            return;
+        }
+        const cgltf_size bufferSize = data->buffer_view->buffer->size;
+        const cgltf_size viewOffset = data->buffer_view->offset;
+        const cgltf_size accessorOffset = data->offset;
+        const cgltf_size totalOffset = viewOffset + accessorOffset;
+
+        if (totalOffset >= bufferSize) {
+            LOG(WARNING) << "Cannot normalize weights, accessor offset exceeds buffer size.";
+            return;
+        }
+
+        const cgltf_size availableBytes = bufferSize - totalOffset;
+        const cgltf_size stride = data->stride;
+
+        cgltf_size maxCount = 0;
+        if (stride > 0 && availableBytes >= sizeof(float4)) {
+            maxCount = 1 + (availableBytes - sizeof(float4)) / stride;
+        }
+
+        cgltf_size safeCount = data->count;
+        if (safeCount > maxCount) {
+            LOG(WARNING) << "Skinning weights accessor count (" << safeCount
+                         << ") exceeds buffer capacity (" << maxCount
+                         << "), clamping to prevent out-of-bounds access.";
+            safeCount = maxCount;
+        }
+
         uint8_t* bytes = (uint8_t*) data->buffer_view->buffer->data;
-        bytes += data->offset + data->buffer_view->offset;
-        for (cgltf_size i = 0, n = data->count; i < n; ++i, bytes += data->stride) {
+        bytes += totalOffset;
+        for (cgltf_size i = 0, n = safeCount; i < n; ++i, bytes += stride) {
             float4* weights = (float4*) bytes;
             const float sum = weights->x + weights->y + weights->z + weights->w;
             *weights /= sum;
@@ -243,8 +275,17 @@ inline void uploadBuffers(FFilamentAsset* asset, Engine& engine,
 
         // For morph targets without buffer_view, use cgltf_accessor_unpack_floats to unpack data directly
         if (!accessor->buffer_view && isMorphTarget) {
-            const size_t floatsCount = accessor->count * cgltf_num_components(accessor->type);
+            const size_t components = cgltf_num_components(accessor->type);
+            if (components > 0 && accessor->count > std::numeric_limits<size_t>::max() / components) {
+                continue;
+            }
+            const size_t floatsCount = accessor->count * components;
+            
+            if (floatsCount > std::numeric_limits<size_t>::max() / sizeof(float)) {
+                continue;
+            }
             const size_t floatsByteCount = sizeof(float) * floatsCount;
+            
             float* floatsData = (float*)malloc(floatsByteCount);
             cgltf_accessor_unpack_floats(accessor, floatsData, floatsCount);
 
@@ -279,8 +320,39 @@ inline void uploadBuffers(FFilamentAsset* asset, Engine& engine,
         }
         if (slot.vertexBuffer) {
             if (utility::requiresConversion(accessor)) {
-                const size_t floatsCount = accessor->count * cgltf_num_components(accessor->type);
+                const cgltf_size bufferSize = accessor->buffer_view->buffer->size;
+                const cgltf_size totalOffset = accessor->buffer_view->offset + accessor->offset;
+                
+                if (totalOffset >= bufferSize) {
+                    continue;
+                }
+                
+                const cgltf_size availableBytes = bufferSize - totalOffset;
+                const cgltf_size stride = accessor->stride;
+                const cgltf_size elementSize = cgltf_calc_size(accessor->type, accessor->component_type);
+                
+                cgltf_size maxCount = 0;
+                if (stride > 0 && availableBytes >= elementSize) {
+                    maxCount = 1 + (availableBytes - elementSize) / stride;
+                }
+                
+                cgltf_size safeCount = accessor->count;
+                if (safeCount > maxCount) {
+                    LOG(WARNING) << "Accessor count exceeds buffer capacity, clamping.";
+                    safeCount = maxCount;
+                }
+
+                const size_t components = cgltf_num_components(accessor->type);
+                if (components > 0 && safeCount > std::numeric_limits<size_t>::max() / components) {
+                    continue;
+                }
+                const size_t floatsCount = safeCount * components;
+                
+                if (floatsCount > std::numeric_limits<size_t>::max() / sizeof(float)) {
+                    continue;
+                }
                 const size_t floatsByteCount = sizeof(float) * floatsCount;
+                
                 float* floatsData = (float*) malloc(floatsByteCount);
                 cgltf_accessor_unpack_floats(accessor, floatsData, floatsCount);
                 BufferObject* bo = BufferObject::Builder().size(floatsByteCount).build(engine);
@@ -317,8 +389,39 @@ inline void uploadBuffers(FFilamentAsset* asset, Engine& engine,
         assert(slot.morphTargetBuffer);
 
         if (utility::requiresPacking(accessor)) {
-            const size_t floatsCount = accessor->count * cgltf_num_components(accessor->type);
+            const cgltf_size bufferSize = accessor->buffer_view->buffer->size;
+            const cgltf_size totalOffset = accessor->buffer_view->offset + accessor->offset;
+            
+            if (totalOffset >= bufferSize) {
+                continue;
+            }
+            
+            const cgltf_size availableBytes = bufferSize - totalOffset;
+            const cgltf_size stride = accessor->stride;
+            const cgltf_size elementSize = cgltf_calc_size(accessor->type, accessor->component_type);
+            
+            cgltf_size maxCount = 0;
+            if (stride > 0 && availableBytes >= elementSize) {
+                maxCount = 1 + (availableBytes - elementSize) / stride;
+            }
+            
+            cgltf_size safeCount = accessor->count;
+            if (safeCount > maxCount) {
+                LOG(WARNING) << "Accessor count exceeds buffer capacity, clamping.";
+                safeCount = maxCount;
+            }
+
+            const size_t components = cgltf_num_components(accessor->type);
+            if (components > 0 && safeCount > std::numeric_limits<size_t>::max() / components) {
+                continue;
+            }
+            const size_t floatsCount = safeCount * components;
+            
+            if (floatsCount > std::numeric_limits<size_t>::max() / sizeof(float)) {
+                continue;
+            }
             const size_t floatsByteCount = sizeof(float) * floatsCount;
+
             float* floatsData = (float*) malloc(floatsByteCount);
             cgltf_accessor_unpack_floats(accessor, floatsData, floatsCount);
             if (accessor->type == cgltf_type_vec3) {