diff --git a/clang/lib/Headers/avx512bf16intrin.h b/clang/lib/Headers/avx512bf16intrin.h
--- a/clang/lib/Headers/avx512bf16intrin.h
+++ b/clang/lib/Headers/avx512bf16intrin.h
@@ -15,7 +15,8 @@
 
 typedef short __m512bh __attribute__((__vector_size__(64), __aligned__(64)));
 typedef short __m256bh __attribute__((__vector_size__(32), __aligned__(32)));
-typedef unsigned short __bfloat16;
+typedef unsigned short __bfloat16
+    __attribute__((deprecated("use unsigned short instead")));
 
 #define __DEFAULT_FN_ATTRS512 \
   __attribute__((__always_inline__, __nodebug__, __target__("avx512bf16"), \
@@ -33,7 +34,7 @@
 ///    A bfloat data.
 /// \returns A float data whose sign field and exponent field keep unchanged,
 ///    and fraction field is extended to 23 bits.
-static __inline__ float __DEFAULT_FN_ATTRS _mm_cvtsbh_ss(__bfloat16 __A) {
+static __inline__ float __DEFAULT_FN_ATTRS _mm_cvtsbh_ss(unsigned short __A) {
   return __builtin_ia32_cvtsbf162ss_32(__A);
 }
 
diff --git a/clang/lib/Headers/avx512vlbf16intrin.h b/clang/lib/Headers/avx512vlbf16intrin.h
--- a/clang/lib/Headers/avx512vlbf16intrin.h
+++ b/clang/lib/Headers/avx512vlbf16intrin.h
@@ -413,7 +413,8 @@
 ///    A float data.
 /// \returns A bf16 data whose sign field and exponent field keep unchanged,
 ///    and fraction field is truncated to 7 bits.
-static __inline__ __bfloat16 __DEFAULT_FN_ATTRS128 _mm_cvtness_sbh(float __A) {
+static __inline__ unsigned short __DEFAULT_FN_ATTRS128
+_mm_cvtness_sbh(float __A) {
   __v4sf __V = {__A, 0, 0, 0};
   __v8hi __R = __builtin_ia32_cvtneps2bf16_128_mask(
       (__v4sf)__V, (__v8hi)_mm_undefined_si128(), (__mmask8)-1);
diff --git a/clang/test/CodeGen/X86/avx512bf16-builtins.c b/clang/test/CodeGen/X86/avx512bf16-builtins.c
--- a/clang/test/CodeGen/X86/avx512bf16-builtins.c
+++ b/clang/test/CodeGen/X86/avx512bf16-builtins.c
@@ -4,7 +4,7 @@
 
 #include <immintrin.h>
 
-float test_mm_cvtsbh_ss(__bfloat16 A) {
+float test_mm_cvtsbh_ss(unsigned short A) {
   // CHECK-LABEL: @test_mm_cvtsbh_ss
   // CHECK: zext i16 %{{.*}} to i32
   // CHECK: shl i32 %{{.*}}, 16
diff --git a/clang/test/CodeGen/X86/avx512bf16-error.c b/clang/test/CodeGen/X86/avx512bf16-error.c
new file mode 100644
--- /dev/null
+++ b/clang/test/CodeGen/X86/avx512bf16-error.c
@@ -0,0 +1,14 @@
+// RUN: %clang_cc1 -fsyntax-only -verify -ffreestanding -triple x86_64-linux-pc %s
+
+// expected-error@+1 3 {{unknown type name '__bfloat16'}}
+__bfloat16 foo(__bfloat16 a, __bfloat16 b) {
+  return a + b;
+}
+
+#include <immintrin.h>
+
+// expected-warning@+2 3 {{'__bfloat16' is deprecated: use unsigned short instead}}
+// expected-note@* 3 {{'__bfloat16' has been explicitly marked deprecated here}}
+__bfloat16 bar(__bfloat16 a, __bfloat16 b) {
+  return a + b;
+}
diff --git a/clang/test/CodeGen/X86/avx512vlbf16-builtins.c b/clang/test/CodeGen/X86/avx512vlbf16-builtins.c
--- a/clang/test/CodeGen/X86/avx512vlbf16-builtins.c
+++ b/clang/test/CodeGen/X86/avx512vlbf16-builtins.c
@@ -162,7 +162,7 @@
   return _mm256_mask_dpbf16_ps(D, U, A, B);
 }
 
-__bfloat16 test_mm_cvtness_sbh(float A) {
+unsigned short test_mm_cvtness_sbh(float A) {
   // CHECK-LABEL: @test_mm_cvtness_sbh
   // CHECK: @llvm.x86.avx512bf16.mask.cvtneps2bf16.128
   // CHECK: ret i16 %{{.*}}
diff --git a/llvm/docs/LangRef.rst b/llvm/docs/LangRef.rst
--- a/llvm/docs/LangRef.rst
+++ b/llvm/docs/LangRef.rst
@@ -3369,8 +3369,8 @@
    * - ``bfloat``
      - 16-bit "brain" floating-point value (7-bit significand).  Provides the
        same number of exponent bits as ``float``, so that it matches its dynamic
-       range, but with greatly reduced precision.  Used in Intel's AVX-512 BF16
-       extensions and Arm's ARMv8.6-A extensions, among others.
+       range, but with greatly reduced precision.  Used in Arm's ARMv8.6-A
+       extensions, among others.
 
    * - ``float``
      - 32-bit floating-point value