diff --git a/agentverse/llms/openai.py b/agentverse/llms/openai.py
index 11b17f02..505a7757 100644
--- a/agentverse/llms/openai.py
+++ b/agentverse/llms/openai.py
@@ -239,6 +239,8 @@ def generate_response(
                     functions=functions,
                     **self.args.dict(),
                 )
+                if not response.choices or response.choices[0].message is None:
+                    raise ValueError("LLM returned empty or filtered response")
 
                 logger.log_prompt(
                     [
@@ -281,6 +283,8 @@ def generate_response(
                     messages=messages,
                     **self.args.dict(),
                 )
+                if not response.choices or response.choices[0].message is None:
+                    raise ValueError("LLM returned empty or filtered response")
                 logger.log_prompt(
                     [
                         {
@@ -335,6 +339,8 @@ async def agenerate_response(
                     functions=functions,
                     **self.args.dict(),
                 )
+                if not response.choices or response.choices[0].message is None:
+                    raise ValueError("LLM returned empty or filtered response")
                 logger.log_prompt(
                     [
                         {
@@ -415,6 +421,8 @@ async def agenerate_response(
                     messages=messages,
                     **self.args.dict(),
                 )
+                if not response.choices or response.choices[0].message is None:
+                    raise ValueError("LLM returned empty or filtered response")
                 self.collect_metrics(response)
                 logger.log_prompt(
                     [
diff --git a/agentverse/memory/chat_history.py b/agentverse/memory/chat_history.py
index 65d00751..6a34b8df 100644
--- a/agentverse/memory/chat_history.py
+++ b/agentverse/memory/chat_history.py
@@ -206,12 +206,15 @@ async def _update_summary_with_batch(
             summary=self.summary, new_events=new_events_batch
         )
 
-        self.summary = await openai_client.chat.completions.acreate(
+        _resp = await openai_client.chat.completions.acreate(
             messages=[{"role": "user", "content": prompt}],
             model=model,
             max_tokens=max_summary_length,
             temperature=0.5,
-        ).choices[0].message.content
+        )
+        if not _resp.choices or _resp.choices[0].message is None:
+            raise ValueError("LLM returned empty or filtered response")
+        self.summary = _resp.choices[0].message.content
 
     def summary_message(self) -> dict:
         return {
diff --git a/scripts/evaluate_responsegen.py b/scripts/evaluate_responsegen.py
index 0b7adabf..16912016 100644
--- a/scripts/evaluate_responsegen.py
+++ b/scripts/evaluate_responsegen.py
@@ -59,6 +59,8 @@ def write_eval_to_file(file, skip=0):
                 time.sleep(min(i**2, 60))
                 continue
             break
+        if not eval_response.choices or eval_response.choices[0].message is None:
+            raise ValueError("LLM returned empty or filtered response")
         text = eval_response.choices[0].message.content
         eval.append(text)
         text = text.replace("\n", "\n\n")