diff --git a/src/core/server/services/comments/moderation/phases/wordList.ts b/src/core/server/services/comments/moderation/phases/wordList.ts
new file mode 100755
index 000000000..71f8eb860
--- /dev/null
+++ b/src/core/server/services/comments/moderation/phases/wordList.ts
@@ -0,0 +1,55 @@
+import {
+  GQLCOMMENT_FLAG_REASON,
+  GQLCOMMENT_STATUS,
+} from "talk-server/graph/tenant/schema/__generated__/types";
+import { ACTION_TYPE } from "talk-server/models/action";
+import {
+  IntermediateModerationPhase,
+  IntermediatePhaseResult,
+} from "talk-server/services/comments/moderation";
+import { containsMatchingPhrase } from "talk-server/services/comments/moderation/wordlist";
+
+// This phase checks the comment against the wordList.
+export const wordList: IntermediateModerationPhase = ({
+  tenant,
+  comment,
+}): IntermediatePhaseResult | void => {
+  // If there isn't a body, there can't be a bad word!
+  if (!comment.body) {
+    return;
+  }
+
+  // Decide the status based on whether or not the current asset/settings
+  // has pre-mod enabled or not. If the comment was rejected based on the
+  // wordList, then reject it, otherwise if the moderation setting is
+  // premod, set it to `premod`.
+  if (containsMatchingPhrase(tenant.wordList.banned, comment.body)) {
+    // Add the flag related to Trust to the comment.
+    return {
+      status: GQLCOMMENT_STATUS.REJECTED,
+      actions: [
+        {
+          action_type: ACTION_TYPE.FLAG,
+          reason: GQLCOMMENT_FLAG_REASON.COMMENT_DETECTED_BANNED_WORD,
+        },
+      ],
+    };
+  }
+
+  // If the comment has a suspect word or a link, we need to add a
+  // flag to it to indicate that it needs to be looked at.
+  // Otherwise just return the new comment.
+
+  // If the wordlist has matched the suspect word filter and we haven't disabled
+  // auto-flagging suspect words, then we should flag the comment!
+  if (containsMatchingPhrase(tenant.wordList.suspect, comment.body)) {
+    return {
+      actions: [
+        {
+          action_type: ACTION_TYPE.FLAG,
+          reason: GQLCOMMENT_FLAG_REASON.COMMENT_DETECTED_SUSPECT_WORD,
+        },
+      ],
+    };
+  }
+};
diff --git a/src/core/server/services/comments/moderation/wordList.spec.ts b/src/core/server/services/comments/moderation/wordList.spec.ts
new file mode 100644
index 000000000..d23705ecf
--- /dev/null
+++ b/src/core/server/services/comments/moderation/wordList.spec.ts
@@ -0,0 +1,46 @@
+import { containsMatchingPhrase } from "talk-server/services/comments/moderation/wordlist";
+
+const phrases = [
+  "cookies",
+  "how to do bad things",
+  "how to do really bad things",
+  "s h i t",
+  "$hit",
+  "p**ch",
+  "p*ch",
+];
+
+describe("containsMatchingPhrase", () => {
+  it("does match on a word in the list", () => {
+    [
+      "how to do really bad things",
+      "what is cookies",
+      "cookies",
+      "COOKIES.",
+      "how to do bad things",
+      "How To do bad things!",
+      "This stuff is $hit!",
+      "That's a p**ch!",
+    ].forEach(word => {
+      expect(containsMatchingPhrase(phrases, word)).toEqual(true);
+    });
+  });
+
+  it("does not match on a word not in the list", () => {
+    [
+      "how to",
+      "cookie",
+      "how to be a great person?",
+      "how to not do really bad things?",
+      "i have $100 dollars.",
+      "I have bad $ hit lling",
+      "That's a p***ch!",
+    ].forEach(word => {
+      expect(containsMatchingPhrase(phrases, word)).toEqual(false);
+    });
+  });
+
+  it("allows an empty list", () => {
+    expect(containsMatchingPhrase([], "test")).toEqual(false);
+  });
+});
diff --git a/src/core/server/services/comments/moderation/wordList.ts b/src/core/server/services/comments/moderation/wordList.ts
new file mode 100644
index 000000000..30cc124ac
--- /dev/null
+++ b/src/core/server/services/comments/moderation/wordList.ts
@@ -0,0 +1,25 @@
+/**
+ * Escape string for special regular expression characters.
+ */
+export function escapeRegExp(str: string) {
+  return str.replace(/[.*+?^${}()|[\]\\]/g, "\\$&"); // $& means the whole matched string
+}
+
+/**
+ * Generate a regular expression that catches the `phrases`.
+ */
+export function generateRegExp(phrases: string[]) {
+  const inner = phrases
+    .map(phrase =>
+      phrase
+        .split(/\s+/)
+        .map(word => escapeRegExp(word))
+        .join('[\\s"?!.]+')
+    )
+    .join("|");
+
+  return new RegExp(`(^|[^\\w])(${inner})(?=[^\\w]|$)`, "iu");
+}
+
+export const containsMatchingPhrase = (phrases: string[], testString: string) =>
+  phrases.length > 0 ? generateRegExp(phrases).test(testString) : false;