diff --git a/.cspell.json b/.cspell.json
index 6bd29ae99dd..79be1a3e99d 100644
--- a/.cspell.json
+++ b/.cspell.json
@@ -405,6 +405,7 @@
         "SHFMT",
         "SNAKEFMT",
         "SOQL",
+        "SOURCEFILEHEADER",
         "SOURCEPATHS",
         "SQLFLUFF",
         "STDLIB",
@@ -1466,4 +1467,4 @@
         "zaach",
         "zricethezav"
     ]
-}
\ No newline at end of file
+}
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 11e0659dee6..16f68f73e32 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -11,6 +11,7 @@ Note: Can be used with `oxsecurity/megalinter@beta` in your GitHub Action mega-l
 - Core
   - mega-linter-runner: Convert to ES6 and upgrade npm dependencies
   - Add rust to checkov as it is a required dependency (to do that, allow to define empty string packages as cargo dependencies in descriptors)
+  - Optimize `@generated` marker scanning ([#2654](https://github.com/oxsecurity/megalinter/pull/2654))
 
 - Media
   - [Achieve Code Consistency: MegaLinter Integration in Azure DevOps](https://techcommunity.microsoft.com/t5/azure-devops-blog/achieve-code-consistency-megalinter-integration-in-azure-devops/ba-p/3939448), by [Don Koning](https://techcommunity.microsoft.com/t5/user/viewprofilepage/user-id/2039143#profile) on [Microsoft Tech Community](https://techcommunity.microsoft.com/)
diff --git a/megalinter/utils.py b/megalinter/utils.py
index 639abceece4..fbf6ee40b68 100644
--- a/megalinter/utils.py
+++ b/megalinter/utils.py
@@ -14,6 +14,8 @@
 from megalinter import config
 from megalinter.constants import DEFAULT_DOCKER_WORKSPACE_DIR
 
+SIZE_MAX_SOURCEFILEHEADER = 1024
+
 REPO_HOME_DEFAULT = (
     DEFAULT_DOCKER_WORKSPACE_DIR
     if os.path.isdir(DEFAULT_DOCKER_WORKSPACE_DIR)
@@ -278,10 +280,9 @@ def file_contains(file_name: str, regex_object: Optional[Pattern[str]]) -> bool:
 
 
 def file_is_generated(file_name: str) -> bool:
-    with open(file_name, "r", encoding="utf-8", errors="ignore") as f:
-        content = f.read()
-    is_generated = "@generated" in content and "@not-generated" not in content
-    return is_generated
+    with open(file_name, "rb") as f:
+        content = f.read(SIZE_MAX_SOURCEFILEHEADER)
+    return b"@generated" in content and b"@not-generated" not in content
 
 
 def decode_utf8(stdout):