diff --git a/studio/frontend/src/features/training/components/hf-dataset-subset-split-selectors.tsx b/studio/frontend/src/features/training/components/hf-dataset-subset-split-selectors.tsx
index d21fd55ff4..148341e4de 100644
--- a/studio/frontend/src/features/training/components/hf-dataset-subset-split-selectors.tsx
+++ b/studio/frontend/src/features/training/components/hf-dataset-subset-split-selectors.tsx
@@ -106,7 +106,7 @@ export function HfDatasetSubsetSplitSelectors({
               : "rounded-lg border border-amber-200 bg-amber-50 px-3.5 py-2.5 text-xs text-amber-700 dark:border-amber-800 dark:bg-amber-950 dark:text-amber-400"
           }
         >
-          Could not fetch dataset splits: {error}
+          {error}
         </div>
       )}
 
diff --git a/studio/frontend/src/features/training/hooks/use-training-actions.ts b/studio/frontend/src/features/training/hooks/use-training-actions.ts
index bfe035cf4a..1f3db35ce5 100644
--- a/studio/frontend/src/features/training/hooks/use-training-actions.ts
+++ b/studio/frontend/src/features/training/hooks/use-training-actions.ts
@@ -16,6 +16,19 @@ const ROLE_REMAP: Record<string, Record<string, string>> = {
   sharegpt: { user: "human", assistant: "gpt", system: "system" },
 };
 
+function normalizeTrainingStartError(message: string): string {
+  const normalized = message.toLowerCase();
+  const isLegacyDatasetScriptError =
+    normalized.includes("failed to check dataset format") &&
+    normalized.includes("dataset scripts are no longer supported");
+
+  if (isLegacyDatasetScriptError) {
+    return "This Hub dataset relies on a legacy custom script and isn’t supported in this training flow.";
+  }
+
+  return message;
+}
+
 export function useTrainingActions() {
   const isStarting = useTrainingRuntimeStore((state) => state.isStarting);
   const startError = useTrainingRuntimeStore((state) => state.startError);
@@ -79,7 +92,9 @@ export function useTrainingActions() {
       const response = await startTraining(payload);
 
       if (response.status === "error") {
-        runtimeStore.setStartError(response.error || response.message);
+        const rawMessage = response.error || response.message;
+        const safeMessage = normalizeTrainingStartError(rawMessage);
+        runtimeStore.setStartError(safeMessage);
         runtimeStore.setStarting(false);
         return false;
       }
@@ -88,9 +103,10 @@ export function useTrainingActions() {
       await syncTrainingRuntimeFromBackend();
       return true;
     } catch (error) {
-      const message =
+      const rawMessage =
         error instanceof Error ? error.message : "Failed to start training";
-      runtimeStore.setStartError(message);
+      const safeMessage = normalizeTrainingStartError(rawMessage);
+      runtimeStore.setStartError(safeMessage);
       runtimeStore.setStarting(false);
       return false;
     }
diff --git a/studio/frontend/src/features/training/stores/training-runtime-store.ts b/studio/frontend/src/features/training/stores/training-runtime-store.ts
index 7d5c86549b..41b0fcca07 100644
--- a/studio/frontend/src/features/training/stores/training-runtime-store.ts
+++ b/studio/frontend/src/features/training/stores/training-runtime-store.ts
@@ -187,7 +187,6 @@ export const useTrainingRuntimeStore = create<TrainingRuntimeStore>()((set) => (
         evalEnabled: payload.eval_enabled ?? state.evalEnabled,
         message: payload.message,
         error: payload.error,
-        startError: null,
         currentStep:
           typeof detailStep === "number" ? Math.max(detailStep, 0) : state.currentStep,
         totalSteps:
diff --git a/studio/frontend/src/hooks/use-hf-dataset-splits.ts b/studio/frontend/src/hooks/use-hf-dataset-splits.ts
index 7b8e4906ec..bc769b5b4b 100644
--- a/studio/frontend/src/hooks/use-hf-dataset-splits.ts
+++ b/studio/frontend/src/hooks/use-hf-dataset-splits.ts
@@ -35,6 +35,37 @@ export interface HfDatasetSplitsResult {
 
 const HF_SPLITS_API = "https://datasets-server.huggingface.co/splits";
 
+function normalizeDatasetSplitsError(message: string): string {
+  const normalized = message.toLowerCase();
+
+  // datasets-server returns technical script/runtime details for legacy datasets.
+  if (
+    normalized.includes("dataset scripts are no longer supported") ||
+    normalized.includes("runs arbitrary python code") ||
+    normalized.includes(".py")
+  ) {
+    return "We can’t load subset/split options for this Hub dataset because it relies on a legacy custom script.";
+  }
+
+  if (
+    normalized.includes("unauthorized") ||
+    normalized.includes("forbidden") ||
+    normalized.includes("access token") ||
+    normalized.includes("private") ||
+    normalized.includes("gated") ||
+    normalized.includes("401") ||
+    normalized.includes("403")
+  ) {
+    return "Unable to load dataset splits. This dataset may be private or gated. Add a Hugging Face token with access and try again.";
+  }
+
+  if (normalized.includes("not found") || normalized.includes("404")) {
+    return "Dataset not found. Check the dataset name and try again.";
+  }
+
+  return "Unable to load dataset split options for this dataset.";
+}
+
 // ---------------------------------------------------------------------------
 // Hook
 // ---------------------------------------------------------------------------
@@ -101,7 +132,18 @@ export function useHfDatasetSplits(
       })
       .catch((err) => {
         if (!controller.signal.aborted) {
-          setError(err.message || "Failed to fetch dataset splits");
+          const rawErrorMessage =
+            err instanceof Error
+              ? err.message
+              : typeof err === "string"
+                ? err
+                : "Failed to fetch dataset splits";
+          console.warn("[useHfDatasetSplits] Failed to fetch dataset splits", {
+            datasetName,
+            message: rawErrorMessage,
+            error: err,
+          });
+          setError(normalizeDatasetSplitsError(rawErrorMessage));
           setEntries([]);
         }
       })