diff --git a/changelog.d/9999-typed-feedback-test-isolation.md b/changelog.d/9999-typed-feedback-test-isolation.md
new file mode 100644
index 0000000000..471e7196b5
--- /dev/null
+++ b/changelog.d/9999-typed-feedback-test-isolation.md
@@ -0,0 +1,3 @@
+### Fixed
+
+- Isolate the default-lane closure-specialization regression from concurrent profiling environment changes and inherited profiling settings.
diff --git a/crates/perry-codegen/tests/typed_feedback.rs b/crates/perry-codegen/tests/typed_feedback.rs
index d4ded4521a..3ee8f733eb 100644
--- a/crates/perry-codegen/tests/typed_feedback.rs
+++ b/crates/perry-codegen/tests/typed_feedback.rs
@@ -1118,6 +1118,11 @@ fn synthetic_arguments_only_method_uses_shape_guarded_direct_call() {
 
 #[test]
 fn typed_feedback_guards_direct_closure_call_specialization() {
+    // This test asserts the non-recording lane. Join the env-mutating tests'
+    // lock and pin that lane, including when the parent enables profiling.
+    let _lock = env_lock();
+    let _feedback = EnvVarGuard::set("PERRY_TYPED_FEEDBACK", None);
+    let _trace = EnvVarGuard::set("PERRY_TYPED_FEEDBACK_TRACE", None);
     let closure_ty = Type::Function(FunctionType {
         params: vec![("x".to_string(), Type::Number, false)],
         return_type: Box::new(Type::Number),