diff --git a/src/model/encoding/__tests__/convertFromHTMLToContentBlocks-test.js b/src/model/encoding/__tests__/convertFromHTMLToContentBlocks-test.js
index db7cd1baab..cd38cb3b03 100644
--- a/src/model/encoding/__tests__/convertFromHTMLToContentBlocks-test.js
+++ b/src/model/encoding/__tests__/convertFromHTMLToContentBlocks-test.js
@@ -50,18 +50,18 @@ describe('convertFromHTMLToContentBlocks', () => {
   ].forEach(tag => testConvertingAdjacentHtmlElementsToContentBlocks(tag));
 
   describe('img tag', function() {
-    test('img with http protocol should have empty content', function() {
+    test('img with http protocol should have camera emoji content', function() {
       const blocks = convertFromHTMLToContentBlocks(
         '<img src="http://www.facebook.com">',
       );
-      expect(blocks.contentBlocks[0].text).toBe(' ');
+      expect(blocks.contentBlocks[0].text).toBe('\ud83d\udcf7');
     });
 
     test('img with data protocol should be correctly parsed', function() {
       const blocks = convertFromHTMLToContentBlocks(
         `<img src="${IMAGE_DATA_URL}">`,
       );
-      expect(blocks.contentBlocks[0].text).toBe(' ');
+      expect(blocks.contentBlocks[0].text).toBe('\ud83d\udcf7');
     });
   });
 });
diff --git a/src/model/encoding/convertFromHTMLToContentBlocks.js b/src/model/encoding/convertFromHTMLToContentBlocks.js
index d92283d953..2b9da2012b 100644
--- a/src/model/encoding/convertFromHTMLToContentBlocks.js
+++ b/src/model/encoding/convertFromHTMLToContentBlocks.js
@@ -405,7 +405,10 @@ function genFragment(
     });
     // Forcing this node to have children because otherwise no entity will be
     // created for this node.
-    node.textContent = ' ';
+    // The child text node cannot just have a space or return as content -
+    // we strip those out.
+    // See https://github.com/facebook/draft-js/issues/231 for some context.
+    node.textContent = '\ud83d\udcf7';
 
     // TODO: update this when we remove DraftEntity entirely
     inEntity = DraftEntity.__create(