diff --git a/README.md b/README.md
index 1979171..29ce1c4 100644
--- a/README.md
+++ b/README.md
@@ -1,6 +1,6 @@
 [![CI](https://github.com/amrali-eg/EncodingChecker/actions/workflows/ci.yml/badge.svg)](https://github.com/amrali-eg/EncodingChecker/actions/workflows/ci.yml)
 
-# EncodingChecker v3.14.4
+# EncodingChecker v3.14.5
 
 EncodingChecker is a Windows tool for finding, checking, and safely converting text-file encodings. Use the GUI for everyday work or the command line for repeatable jobs.
 
diff --git a/docs/RELEASE-NOTES-v3.14.5.md b/docs/RELEASE-NOTES-v3.14.5.md
new file mode 100644
index 0000000..7f38928
--- /dev/null
+++ b/docs/RELEASE-NOTES-v3.14.5.md
@@ -0,0 +1,55 @@
+# EncodingChecker v3.14.5
+
+Three fixes to what EC records and writes around a conversion. What EC detects,
+converts and refuses is unchanged; the fixes are in how a file that could not be
+processed is planned, how a failed snapshot is recorded, and how the GUI saves reports.
+
+## What changed
+
+- **A file that fails full validation is planned as a refusal.** A file that looked
+  like the target codec in the 64 KiB detection sample but held invalid bytes later was
+  reported as an error, yet its planned action stayed `Unchanged`. The plan summary and
+  the confirmation dialog counted it under "Already in the target encoding" and said "No
+  conversion is needed". It is now planned as `Refuse`, so it appears under "Cannot be
+  processed safely". Nothing about writing changed: the file was never converted, the CLI
+  already exited 3, and applying the plan still re-reads the file, reports the failure
+  with its reason and leaves the bytes untouched.
+- **The GUI text and CSV exports no longer erase an existing report when the write
+  fails.** They opened the chosen file directly, which truncates it first. They now stage
+  the report in a temporary file and install it only when complete, as the CLI reports,
+  journals, plans and settings already did. A read-only report, or a link, is refused
+  with a message rather than replaced.
+- **A failed snapshot no longer keeps an earlier run's hash.** GUI rows survive between
+  runs. A row whose source could not be read when a plan was prepared was refused as an
+  error but kept the hash and size from its previous snapshot, which the plan and the
+  journal could then record as if they described this attempt. Both are now cleared.
+  A stale hash would have failed the check at apply, so the effect was a wrong recorded
+  value, not an unsafe write.
+- Documentation and comments only: `SAFETY.md` now says that a failure to mark the
+  recovery record complete happens after installation and leaves the verified output in
+  place, and names `utf-32le`/`utf-32be` for BOM-less UTF-32; two code comments that
+  overstated what an unchanged file and a missing detected codec mean were corrected.
+
+## Compatibility
+
+Conversion semantics stay at **7**, the plan schema at **6**, the journal schema at
+**6**. Plans and journals from any 3.14.x release still apply unchanged.
+
+One difference to expect: a plan written by an earlier 3.14.x release for a file of the
+first kind records `Unchanged`, so its confirmation dialog still lists that file as
+already in the target encoding. Applying it still fails that file safely. Plans written by
+this release record `Refuse`, and the journal's planned action for such a file is now
+`Refuse` as well.
+
+## Verification
+
+- 933 tests pass, none skipped; release build with no warnings
+- The GUI smoke suite passes all ten phases against an ordinary Release build of this
+  release commit (report `Outcome` Passed, no evidence errors)
+- `docs/Test-DefectBacklog.ps1` passes
+- The **Shared Unicode detector parity** check is unaffected: neither shared detector file
+  changed
+- The four-corpus audit was **not** run. It applies to detection and conversion-policy
+  changes; `ConversionPolicy` decides exactly as before, and the only behavioural changes
+  are the planned label of a file that failed validation and the export and snapshot
+  handling above.
diff --git a/sources/EncodingChecker/Properties/AssemblyInfo.cs b/sources/EncodingChecker/Properties/AssemblyInfo.cs
index ad8e12a..c60ff9a 100644
--- a/sources/EncodingChecker/Properties/AssemblyInfo.cs
+++ b/sources/EncodingChecker/Properties/AssemblyInfo.cs
@@ -41,5 +41,5 @@
 // You can specify all the values, or you can default the Build and Revision Numbers
 // by using the '*' as shown below:
 // [assembly: AssemblyVersion("1.0.*")]
-[assembly: AssemblyVersion("3.14.4.0")]
-[assembly: AssemblyFileVersion("3.14.4.0")]
+[assembly: AssemblyVersion("3.14.5.0")]
+[assembly: AssemblyFileVersion("3.14.5.0")]