diff --git a/server/models/Book.js b/server/models/Book.js
index 19923750e..df2014c4e 100644
--- a/server/models/Book.js
+++ b/server/models/Book.js
@@ -439,6 +439,7 @@ class Book extends Model {
    */
   async updateAuthorsFromRequest(authors, libraryId) {
     if (!Array.isArray(authors)) return null
+    authors = parseNameString.dedupeNames(authors)
 
     if (!this.authors) {
       throw new Error(`[Book] Cannot update authors because authors are not loaded for book ${this.id}`)
@@ -466,6 +467,8 @@ class Book extends Model {
     const authorsAdded = []
     for (const authorName of newAuthorNames) {
       const { author, created } = await authorModel.findOrCreateByNameAndLibrary(authorName, libraryId)
+      // findOrCreate matches case-insensitively, so a second spelling can resolve to an author just linked
+      if (this.authors.some((au) => au.id === author.id)) continue
       await bookAuthorModel.create({ bookId: this.id, authorId: author.id })
       if (created) {
         SocketAuthority.emitter('author_added', author.toOldJSON())
diff --git a/server/scanner/BookScanner.js b/server/scanner/BookScanner.js
index 2b106cdc8..f8762a1cf 100644
--- a/server/scanner/BookScanner.js
+++ b/server/scanner/BookScanner.js
@@ -703,6 +703,9 @@ class BookScanner {
       bookMetadata.description = htmlSanitizer.sanitize(bookMetadata.description)
     }
 
+    // One dedupe after every metadata source. A repeated name inserts the same bookAuthor twice and aborts the scan.
+    bookMetadata.authors = parseNameString.dedupeNames(bookMetadata.authors)
+
     return bookMetadata
   }