diff --git a/llvm/docs/DeveloperPolicy.rst b/llvm/docs/DeveloperPolicy.rst
--- a/llvm/docs/DeveloperPolicy.rst
+++ b/llvm/docs/DeveloperPolicy.rst
@@ -570,9 +570,16 @@
 of reasons.
 
 For these reasons, new targets are *always* added as *experimental* until
-they can be proven stable, and later moved to non-experimental. The difference
-between both classes is that experimental targets are not built by default
-(need to be added to -DLLVM_TARGETS_TO_BUILD at CMake time).
+they can be proven stable, and later moved to non-experimental. The differences
+between both classes are:
+
+* Experimental targets are not built by default (need to be added to
+  -DLLVM_TARGETS_TO_BUILD at CMake time).
+
+* It is solely the responsibility of the community behind the target to maintain
+  that target. This means that it is not appropriate to revert a change
+  unrelated to the target or ask for something in it to be fixed, if that change
+  only causes a problem when the experimental target is enabled.
 
 The basic rules for a back-end to be upstreamed in **experimental** mode are: