What content can ShanHaiYin analyze?
ShanHaiYin supports images, text, audio, and video. Each content type uses a dedicated analysis workflow, so their results are not mixed.
How should I read a detection result?
Start with the primary finding and estimated AI likelihood, then review provenance credentials, generation indicators, and other supporting evidence. The request ID traces the service call, while the SHA-256 fingerprint links the report to the analyzed content.
Are uploaded files stored permanently?
Uploads are processed temporarily and sent to the relevant cloud analysis service. Temporary files are not published or reused as public content. Account history stores detection records and reports that you choose to save. See the Privacy Policy for details.
Why can an AI image receive a very low score?
Screenshots, heavy compression, cropping, or extensive editing can weaken detectable indicators. Use the original file when possible and treat the result as technical evidence rather than a final legal determination.
What is the SHA-256 fingerprint for?
It is a unique fingerprint calculated from the analyzed content. If the content changes, the fingerprint normally changes, helping you identify a mismatch between a report and a file.
How does text detection work?
The text model evaluates structural and stylistic generation indicators and returns a finding with confidence information. Complete passages between 350 and 2,000 characters generally produce more stable evidence than short fragments.
Why should I upload the original audio or video file?
Audio analysis needs continuous sound, and video analysis needs continuous frames and changes over time. Short clips, screenshots, and repeated transcoding may remove useful evidence.
How are points charged?
Image and text analysis cost 1 point per request. Audio costs 3 points per started minute and video costs 8 points per started minute. Any partial minute is rounded up to a full minute.