Thanks for the excellent report @tigerjack - I've tracked this down and I beleive this is rclone's bug, introduced in the v1.75.0 betas after 2026-07-13 and included in v1.75.0. Sorry about that!
First the important bit: your data is safe. The file content is uploaded correctly and rclone can read everything back (which is why your sync shows the correct sha1) - it is only the Proton apps which refuse to open the files.
The background: Proton is migrating Drive to a new generation of OpenPGP encryption (their announcement) and rclone v1.75.0 added support for uploading in the new format, with spec guidance kindly provided by a Proton Drive developer in this thread. The file content encryption is correct. However two auxiliary pieces - the encrypted metadata (modification time etc) and the per-block signatures - are required by Proton's clients to stay in the old format. The OpenPGP library rclone uses silently upgrades those to the new format too, because the file's new-style key advertises support for it. The official apps then fail with "the file or some of its data cannot be decrypted", while rclone (whose library reads both formats) carries on happily.
What's affected: only files newly created on the remote with v1.75.0 or a beta from after 2026-07-13. Files uploaded with v1.74.4 or earlier are untouched, as are updates to such files.
I'm working on a fix now and will post a link to a beta in this thread when it is ready. Once you are on the fixed version, re-uploading the affected files (delete them on the remote and copy again, or use rclone copy --ignore-times - noting that re-uploads everything in the path) will make them readable in the web app again. If you need the web app in the meantime, v1.74.4 uploads in the old format.
The context canceled NOTICE/ERROR messages at the end of your log are unrelated by the way - that's just noise from connections being torn down as the sync finishes.
On the wider discussion above: Proton have actually been actively helpful recently - the new format support only exists because one of their Drive developers gave us detailed guidance on the forum - and this particular bug is entirely ours.