Start here: an on-site photo is a claim, not the whole record
Many disputes do not start with a contract paragraph. They start with a photo: a contractor sends a “completed repair” image, a tenant sends property-damage photos, a customer sends a broken-product picture, an employee sends a field-visit snapshot, or a claimant sends a scene photo to support reimbursement or compensation.
The hard part is that you are not there. The image may look plausible, but you cannot tell whether it is current, cropped, reused, edited, staged, or AI-generated.
Do not open by accusing the other person. Preserve the file, run a technical check when the decision matters, and ask for follow-up evidence that is harder to fake.
When should you verify an on-site photo?
Not every casual image needs forensic review. Verification is worth the effort when the image affects money, liability, job completion, reimbursement, or trust.
Contractor completion and repair work
A contractor says a roof patch, HVAC repair, plumbing job, appliance repair, cleaning job, or installation is complete and asks for payment. If you cannot inspect the site immediately, the photo should be treated as one piece of evidence—not as final sign-off.
Rental deposits and property damage
Landlords and tenants often rely on photos when arguing about damage, cleaning, furniture, flooring, walls, appliances, or deposit deductions. The file’s source, time, continuity, and surrounding records matter.
Shipping damage, product claims, and refunds
A customer sends images of a cracked item, leaking package, stain, mold, missing part, or damaged box. The seller should preserve the original file, compare it with the order and shipment record, and request video or additional angles when the claim affects a refund or replacement.
Field-service, merchandising, and attendance proof
A field employee or subcontractor sends a storefront, equipment, shelf, delivery, or job-site image as proof of attendance. If the photo affects pay, commission, compliance, or a customer promise, require more than a single image when something looks off.
Insurance-style or reimbursement claims
Property, vehicle, workplace, and expense claims often include photos. The National Insurance Crime Bureau discusses staged accidents as a type of insurance fraud, and consumer agencies regularly warn about repair and rental scams. Image checks are not a replacement for investigation, but they help document whether a file itself looks reliable.
Step 1: Preserve the original image or video
Save the file exactly as received. If it came through email, a marketplace dashboard, a chat app, a shared drive, or a contractor portal, keep the original download and the message context.
Before analysis, avoid:
- cropping or resizing the image;
- drawing arrows, boxes, or circles on the original;
- forwarding it through another app that may compress it;
- taking a screenshot of the photo instead of saving the file;
- changing brightness, color, or format.
If you need a marked-up version for internal discussion, make a copy and label it as an annotation. Keep the original untouched.
Step 2: Use ShanHaiYin to check for AI, editing, and provenance signals
Upload the preserved file to ShanHaiYin AI image and video analysis. ShanHaiYin checks for AI-generation risk, deepfake or assisted-editing signals, available provenance or file labels, and image-level inconsistencies such as unusual textures, edges, lighting, or local edits.
The downloadable report is not a court ruling, an insurance decision, or a final fraud finding. It is a technical record that helps you say: this file has indicators that should be reviewed before payment, refund, reimbursement, deposit withholding, or sign-off.
That distinction is useful. A job record, payment dashboard, work order, claim file, or rental inspection record answers the business question. Image analysis answers whether the photo or video itself appears technically reliable.
Step 3: Ask for follow-up evidence that is harder to fake
If the report flags risk, or if the file already looks suspicious, do not argue over one image forever. Ask for stronger evidence.
- a continuous video from wide shot to close-up;
- a new photo showing today’s date, project number, room number, order number, or asset ID;
- another angle of the same damage or completed work;
- before-and-after photos for repairs, cleaning, or installation;
- packaging, label, tracking number, full item, and close-up damage for shipment claims;
- a short video showing equipment running, lights on, water flow, serial number, or the surrounding site.
Single images are easier to reuse, crop, edit, or generate. Continuous context is harder to fake and easier for a human reviewer to evaluate.
Copy-ready wording
Customer damage claim
Thanks for sending the photos. Because these images affect refund or replacement handling, we need to verify the evidence before making a decision. We preserved the original file and ran an image check with ShanHaiYin. The report indicates: [quote the actual finding only]. Please send a continuous video showing the packaging, label, full item, and damaged area without cuts.
Contractor or field-service completion
We received the site photos. Since they will be used for completion approval and payment, we need one more verification step. Please send a continuous site video from the overall area to the completed work, including today’s date or the project number. We will review sign-off after the materials are complete.
Tenant, landlord, or property manager
We received the property photos. Because they may affect deposit, repair, or responsibility decisions, please provide a continuous video and additional angles showing the room, item, or area in context. We are preserving the original files and will review them together with the inspection record.
Marketplace, platform, insurer, or internal reviewer
The other party submitted on-site images as evidence for [refund / claim / payment / deposit / completion]. We preserved the original files without editing and generated a technical image/video analysis report with ShanHaiYin. The report indicates: [quote the actual finding only]. Please review the original files, the report, and the related order, contract, inspection, or job record before making a final decision.
Minimal evidence package
You do not need to build a legal brief for every dispute. Start with three things:
- the original photo or video the other party submitted;
- the full ShanHaiYin report;
- the order, invoice, contract, claim, inspection, job, room, or asset record that connects the file to the matter.
This package explains what was submitted, what technical concerns were found, and which transaction or project it relates to. If the matter escalates, you can add more evidence later.
What not to do
- Do not fabricate or alter evidence yourself.
- Do not edit the original image before analysis.
- Do not accuse the other party of fraud based on a tool result alone.
- Do not describe a detection report as a court-grade forensic conclusion.
- Do not ignore business records: contracts, invoices, payment status, inspection notes, delivery records, and site follow-ups still matter.
FAQ
Can an on-site photo be used as evidence by itself?
It can be part of the record, but it should rarely be the only evidence for payment, reimbursement, refund, deposit withholding, or claim approval. Match it with the original file, follow-up video, job records, timestamps, order records, and human review.
Can ShanHaiYin analyze video as well as images?
Yes. ShanHaiYin supports image, text, audio, and video analysis. For contractor completion, property damage, field-service visits, or insurance-style disputes, video often provides stronger context than a single image.
Does a detection report prove someone committed fraud?
No. A detection report can document AI-generation, editing, or manipulation risk in the file. It should be used as supporting technical evidence alongside contracts, work orders, original files, payment records, and manual review.
What if the other party refuses to send a follow-up video?
State that the current image is not enough to approve payment, reimbursement, refund, deposit withholding, or claim handling. Ask for a continuous video, a dated photo, a project number, a room or asset ID, or another verifiable reference. Preserve the refusal as part of the record.
Should I blur or mark up the image before analysis?
Use the original file whenever possible. Cropping, blurring, compression, screenshots, and markups can remove useful signals. If you need an annotated copy, keep it separate from the original file and the detection report.
Sources and further reading
Bottom line
When a photo or video determines money, responsibility, or sign-off, do not rely on gut feeling and do not argue over one image forever. Preserve the original file, run a technical check with ShanHaiYin when the stakes justify it, and ask for follow-up evidence that connects the file to the real site.
The goal is not to turn an AI detector into a judge. The goal is to move from “this looks wrong” to “this file has technical questions, so we need a better record before deciding.”