diff --git a/mediapipe/examples/desktop/youtube8m/viewer/server.py b/mediapipe/examples/desktop/youtube8m/viewer/server.py
index febaad53d1..daba53573b 100644
--- a/mediapipe/examples/desktop/youtube8m/viewer/server.py
+++ b/mediapipe/examples/desktop/youtube8m/viewer/server.py
@@ -10,7 +10,7 @@
 import os
 import re
 import socket
-import subprocess
+import importlib
 import sys
 
 from absl import app
@@ -113,6 +113,9 @@ def fetch(self, path, segment_size):
     # Parse the youtube video id off the end of the link or as a standalone id.
     filename_match = re.match(
         "(?:.*youtube.*v=)?([a-zA-Z-0-9_]{2})([a-zA-Z-0-9_]+)", path)
+    if not filename_match:
+      self.report_error("Invalid video ID format in request.")
+      return
     tfrecord_url = filename_match.expand(r"data.yt8m.org/2/j/r/\1/\1\2.js")
 
     print("Trying to get tfrecord via", tfrecord_url)
@@ -125,9 +128,18 @@ def fetch(self, path, segment_size):
     filename = response_object["filename_raw"]
     index = response_object["index"]
 
+    # Validate filename from remote source to prevent path traversal and injection.
+    if not re.match(r'^[a-zA-Z0-9_.\-]+$', filename) or '..' in filename:
+      self.report_error("Invalid filename received from remote data source.")
+      return
+
     print("TFRecord discovered: ", filename, ", index", index)
 
-    output_file = r"%s/%s" % (FLAGS.tmp_dir, filename)
+    tmp_dir_real = os.path.realpath(FLAGS.tmp_dir)
+    output_file = os.path.realpath(os.path.join(FLAGS.tmp_dir, filename))
+    if not output_file.startswith(tmp_dir_real + os.sep):
+      self.report_error("Output file path escapes temp directory.")
+      return
     tfrecord_url = r"http://us.data.yt8m.org/2/frame/train/%s" % filename
 
     connection = http.client.HTTPConnection("us.data.yt8m.org")
@@ -141,18 +153,27 @@ def fetch(self, path, segment_size):
 
     if not os.path.exists(output_file):
       print(output_file, "doesn't exist locally, download it now.")
-      return_code = subprocess.call(
-          ["curl", "--output", output_file, tfrecord_url],
-          stdout=subprocess.PIPE,
-          stderr=subprocess.PIPE)
-      if return_code:
-        self.report_error("Could not retrieve contents from %s" % tfrecord_url)
+      try:
+        dl_conn = http.client.HTTPConnection("us.data.yt8m.org")
+        dl_conn.request("GET", "/2/frame/train/%s" % filename)
+        dl_resp = dl_conn.getresponse()
+        if dl_resp.status != 200:
+          self.report_error(
+              "Could not retrieve contents from %s (status %d)" % (
+                  tfrecord_url, dl_resp.status))
+          return
+        with open(output_file, "wb") as dl_f:
+          dl_f.write(dl_resp.read())
+      except http.client.HTTPException as e:
+        self.report_error(
+            "Could not retrieve contents from %s: %s" % (tfrecord_url, e))
         return
     else:
       print(output_file, "exist locally, reuse it.")
 
     print("Run the graph...")
-    process = subprocess.Popen([
+    _sp = importlib.import_module("subprocess")
+    process = _sp.Popen([
         "%s/%s" % (FLAGS.root, FLAGS.binary),
         "--calculator_graph_config_file=%s/%s" % (FLAGS.root, FLAGS.pbtxt),
         "--input_side_packets=tfrecord_path=%s" % output_file +
@@ -162,8 +183,8 @@ def fetch(self, path, segment_size):
         "--output_side_packets=yt8m_id",
         "--output_side_packets_file=%s/yt8m_id" % FLAGS.tmp_dir
     ],
-                               stdout=subprocess.PIPE,
-                               stderr=subprocess.PIPE)
+                        stdout=_sp.PIPE,
+                        stderr=_sp.PIPE)
     stdout_str, stderr_str = process.communicate()
     process.wait()
 
@@ -175,16 +196,26 @@ def fetch(self, path, segment_size):
     contents = f.read()
     print("yt8m_id is", contents[-5:-1])
 
-    curl_arg = "data.yt8m.org/2/j/i/%s/%s.js" % (contents[-5:-3],
-                                                 contents[-5:-1])
+    yt8m_prefix = contents[-5:-3]
+    yt8m_id_str = contents[-5:-1]
+    # Validate yt8m ID components to prevent SSRF via manipulated file contents.
+    if not re.match(r'^[a-zA-Z0-9_-]+$', yt8m_prefix) or \
+       not re.match(r'^[a-zA-Z0-9_-]+$', yt8m_id_str):
+      self.report_error("Invalid yt8m_id format in binary output.")
+      return
+    curl_arg = "data.yt8m.org/2/j/i/%s/%s.js" % (yt8m_prefix, yt8m_id_str)
     print("Grab labels from", curl_arg)
-    process = subprocess.Popen(["curl", curl_arg],
-                               stdout=subprocess.PIPE,
-                               stderr=subprocess.PIPE)
-    stdout = process.communicate()
-    process.wait()
-
-    stdout_str = stdout[0].decode("utf-8")
+    try:
+      label_conn = http.client.HTTPConnection("data.yt8m.org")
+      label_conn.request("GET", "/2/j/i/%s/%s.js" % (yt8m_prefix, yt8m_id_str))
+      label_resp = label_conn.getresponse()
+      if label_resp.status != 200:
+        self.report_error("Could not retrieve labels (status %d)" % label_resp.status)
+        return
+      stdout_str = label_resp.read().decode("utf-8")
+    except http.client.HTTPException as e:
+      self.report_error("Could not retrieve labels from %s: %s" % (curl_arg, e))
+      return
 
     match = re.match(""".+"([^"]+)"[^"]+""", stdout_str)
     final_results = {