diff --git a/.reuse/dep5 b/.reuse/dep5
index 865db41d..d1031d67 100644
--- a/.reuse/dep5
+++ b/.reuse/dep5
@@ -70,6 +70,7 @@ Files: tests/resources/linn.png
  tests/resources/ccitt.pdf
  tests/resources/cardinal.pdf
  tests/resources/jbig2.pdf
+ tests/resources/jbig2_baddevicen.pdf
  tests/resources/skew.pdf
  tests/resources/rotated_skew.pdf
  tests/resources/poster.pdf
diff --git a/src/ocrmypdf/_exec/ghostscript.py b/src/ocrmypdf/_exec/ghostscript.py
index 7f14940f..78b5434d 100644
--- a/src/ocrmypdf/_exec/ghostscript.py
+++ b/src/ocrmypdf/_exec/ghostscript.py
@@ -17,7 +17,7 @@ from subprocess import PIPE, CalledProcessError
 from packaging.version import Version
 from PIL import Image, UnidentifiedImageError
 
-from ocrmypdf.exceptions import SubprocessOutputError
+from ocrmypdf.exceptions import ColorConversionNeededError, SubprocessOutputError
 from ocrmypdf.helpers import Resolution
 from ocrmypdf.subprocess import get_version, run, run_polling_stderr
 
@@ -64,10 +64,6 @@ class DuplicateFilter(logging.Filter):
 log.addFilter(DuplicateFilter(log))
 
 
-# Ghostscript executable - gswin32c is not supported
-GS = 'gswin64c' if os.name == 'nt' else 'gs'
-
-
 def version() -> Version:
     return Version(get_version(GS))
 
@@ -77,6 +73,15 @@ def _gs_error_reported(stream) -> bool:
     return bool(match)
 
 
+def _gs_devicen_reported(stream) -> bool:
+    match = re.search(
+        r'DeviceN.*inappropriate alternate',
+        stream,
+        flags=re.IGNORECASE | re.MULTILINE,
+    )
+    return bool(match)
+
+
 def rasterize_pdf(
     input_file: os.PathLike,
     output_file: os.PathLike,
@@ -250,7 +255,6 @@ def generate_pdfa(
         ]
     )
     args_gs.extend(fspath(s) for s in pdf_pages)  # Stringify Path objs
-
     try:
         with Path(output_file).open('wb') as output:
             p = run_polling_stderr(
@@ -279,3 +283,5 @@ def generate_pdfa(
             # the **** pattern to split the stderr into parts.
             for part in stderr.split('****'):
                 log.error(part)
+        if _gs_devicen_reported(stderr):
+            raise ColorConversionNeededError()
diff --git a/src/ocrmypdf/exceptions.py b/src/ocrmypdf/exceptions.py
index 6f549286..fd221e87 100644
--- a/src/ocrmypdf/exceptions.py
+++ b/src/ocrmypdf/exceptions.py
@@ -137,3 +137,16 @@ class TaggedPDFError(InputFileError):
         override this error.
         """
     )
+
+
+class ColorConversionNeededError(BadArgsError):
+    """PDF needs color conversion."""
+
+    message = dedent(
+        """\
+        The input PDF has an unusual color space. Use
+        --color-conversion-strategy to convert to a common color space
+        such as RGB, or use --output-type pdf to skip PDF/A conversion
+        and retain the original color space.
+        """
+    )
diff --git a/tests/resources/jbig2_baddevicen.pdf b/tests/resources/jbig2_baddevicen.pdf
new file mode 100644
index 00000000..8653426d
Binary files /dev/null and b/tests/resources/jbig2_baddevicen.pdf differ
diff --git a/tests/test_ghostscript.py b/tests/test_ghostscript.py
index 281c80ac..990192c5 100644
--- a/tests/test_ghostscript.py
+++ b/tests/test_ghostscript.py
@@ -13,7 +13,7 @@ import pytest
 from PIL import Image, UnidentifiedImageError
 
 from ocrmypdf._exec.ghostscript import DuplicateFilter, rasterize_pdf
-from ocrmypdf.exceptions import ExitCode
+from ocrmypdf.exceptions import ColorConversionNeededError, ExitCode
 from ocrmypdf.helpers import Resolution
 
 from .conftest import check_ocrmypdf, run_ocrmypdf_api
@@ -126,6 +126,16 @@ def test_ghostscript_feature_elision(resources, outpdf):
     )
 
 
+def test_ghostscript_mandatory_color_conversion(resources, outpdf):
+    with pytest.raises(ColorConversionNeededError):
+        check_ocrmypdf(
+            resources / 'jbig2_baddevicen.pdf',
+            outpdf,
+            '--plugin',
+            'tests/plugins/tesseract_noop.py',
+        )
+
+
 def test_rasterize_pdf_errors(resources, no_outpdf, caplog):
     with patch('ocrmypdf._exec.ghostscript.run') as mock:
         # ghostscript can produce