Earlier quoted context omitted.
May I ask you why? I use and have used both PUT presigned urls and POST signed policies for uploading file to S3 without too many problems, but your remark worries me a bit. Am I missing something?
We get our customers to upload to S3 directly so we don't have to handle their ingress traffic through our infra. Problems involve mostly when the client cocks up as it's near impossible to distinguish between broken files and good ones. This leads to the assumption that the clients have uploaded a file and they haven't.
But if you’re saying clients will upload the wrong file, then that’s not really AWS’s fault is it?
If you’re saying the file itself is corrupted, I’m not sure how? The upload should fail if the content length is incorrect, and s3 doesn’t make the contents visible unless the upload completes.