Fact · Voice privacy and security
The openai-whisper loader takes the expected SHA256 checksum from the download address of the model, then compares it with the hash of the file saved on disk.
Checked against its source on October 17, 2026
In the source's own words
“expected_sha256 = url.split("/")[-2] ... if hashlib.sha256(model_bytes).hexdigest() == expected_sha256:”
openai/whisper: whisper/__init__.py (model download and loading code), OpenAI, GitHub (openai/whisper, main branch), 2026, _download function in whisper/__init__.py.
Open the original- What it measures
- The loader keeps a dictionary of model download addresses in the same file. Each address contains the model's checksum as a path segment, and the loader splits that segment out of the address before it compares the hashes.
- What it does not show
- The checksum is written into the package you installed. The comparison confirms that a file matches that package's checksum, so it catches corruption or a changed download, but it cannot show that the package itself is authentic.