Koncile detects tampering in PDF and image files, and certificates generated or edited by AI, then returns a fraud score from 0 to 100: catch a filled in template or an edited copy before a death claim is paid.
Create a free account to run the analysis.
Create a free accountFor document volumes or an API, talk to an expert.
The document checker used by insurers, banks and large companies
4.8 / 5
Start with the copy itself: a real death certificate is a certified copy issued by a state or county vital records office, with a seal, a registrar's signature and a state file number. Then check that names and dates agree with the claim. Finally check the file, which the eye cannot: the software that produced the PDF, its dates and its editing traces.
Certified copies, scans and PDFs: the checker runs four levels of checks. A filled in template or a replaced date leaves traces, either in the computer file or in the pixels.
A certified copy is issued by a state or county vital records office, then scanned or sent as a PDF. Koncile reads which software produced the file. A form builder, a template site or a word processor is not how a registrar issues a certificate.
The most profitable edit is a real certificate with another name or another date. An edited PDF often keeps its earlier versions: Koncile rebuilds them and shows the before and the after, even when seal and layout are perfect.
On a scan or a photo, K-Detect, our image analysis model, isolates edited zones: a date retyped over the original, a seal or a signature pasted in. It also flags certificates generated in whole or in part by AI.
The decedent's name and date of birth must match the policy and the ID document, and the dates on the certificate must fit the claim. Koncile flags every mismatch; your team confirms the record with the issuing office.
In the US, each state issues its own certificate, built on a standard federal model. In the UK, the death certificate is a certified copy of the entry in the register of deaths. Knowing what should be there is the first check.
Legal name, sex, date and place of birth, last residence, marital status and parents' names, then the place and date of death. A life insurance claim checks these fields first, against the policy and the insured's ID document.
The physician, medical examiner or coroner signs the medical part: cause of death, manner of death, date pronounced dead. Some states issue copies without the cause of death. A missing certifier, or a certifier whose role does not fit the circumstances, deserves a question.
The registrar files the record and gives it a state file number and a date filed. The copy you receive is issued later by the vital records office, with its own issue date and seal. Some states also issue informational copies, marked as not valid to establish identity: they cannot support a claim.
A death certificate opens a payout: a life insurance benefit, a group policy, an estate, the write off of a debt. It is checked by fewer people than an ID or a bank statement, and claims teams rarely see two from the same office. Four cases come up.
Form builders, PDF form sites and template galleries rank on the first page of Google for death certificates, and a chatbot prompt produces one in minutes. These files never went through a vital records system: a consumer tool in the metadata, a creation date after the date of issue, no file number in the state's format.
A genuine certificate from another death, with the name, the date of birth or the date of death replaced in a PDF editor. Seal, layout and signatures stay in place. What gives it away: an editing programme in the metadata, a file rewritten after its issue date, the old value still inside the PDF.
A photo or scan of a certificate where a field was regenerated by an image model, or a whole certificate generated from a description. On an image, the original metadata is gone: image analysis looks for edited zones and generated areas instead.
A death that occurred in another country comes with a foreign certificate, in another language and format, from a registry the claims team cannot easily query. It is the hardest case to verify, and the file check is often the only quick signal available.
This page is about detection only. Koncile's checker is for businesses verifying the documents they receive, never for producing one.
Without a tool, these three checks stop clumsy fakes. Ask for the certified copy, or the PDF exactly as the family or the funeral home received it, not a screenshot.
The issuing office must be the vital records office of the place of death, the copy must carry its seal and a registrar's signature, and the state file number must be present. An informational copy, or a copy with no issuing office, does not support a claim.
Date of death, date pronounced dead, date filed and date issued must follow each other. The decedent's name, date of birth and last address must match the policy, the ID document and the claim form.
Use our free PDF metadata viewer: a form builder, a word processor or a PDF editor in Creator or Producer is a red flag. When the amount justifies it, confirm the record directly with the issuing vital records office.
Koncile's fraud detection API gives you a score from 0 to 100 and detailed explanations of manipulation risks, right inside your claims system.
Decedent, dates, certifier, file number and issuing office, ready to cross check.
High, medium or low level, with a one sentence explanation.
High risk from 80.
A one line summary ready for your claims system.
curl -X POST https://api.koncile.ai/v1/upload_file \ -H "Authorization: Bearer $API_KEY" \ -F "files=@death-certificate.pdf"requests.post("https://api.koncile.ai/v1/upload_file/", headers={"Authorization": f"Bearer {API_KEY}"}, files=[("files", open("death-certificate.pdf", "rb"))]){ "status": "DONE", "General_fields": { "Issuing_office": { "value": "State vital records office", "confidence_score": 0.97 }, "Date_of_death": {"value": "2026-03-14", "confidence_score": 0.98} }, "Fraud_detection": { "risk_score": 87, "summary": "PDF filled in from an online template, no issuing system trace", "signals": [ {"name": "Generator fingerprint", "severity": "high"}, {"name": "Editing software", "severity": "medium"} ] }}Wherever a death certificate releases money, a fake is costly and rarely caught after payment. Here is where the checker fits in.
The certificate is checked on receipt, with the claim form, the beneficiary's ID and the policy. Anomalies are flagged before the benefit is paid, with a score the handler can explain.
Employers' group policies, pension funds and annuity providers stop or transfer payments on a death certificate. The check runs before a survivor benefit starts or a payment stops.
A death certificate closes accounts, releases funds to heirs and ends a loan or a card debt. The certificate is scored and its names cross checked with the account holder before funds move.
There is no reliable public figure on forged death certificates alone. The closest sourced figures come from French insurers, and from our own analyses.
In French protection insurance, false declarations come first and forged documents right behind. One lesson from our own tests: a genuine civil status record once came out high risk, because it had been through several layers and successive edits before reaching us. Civil records are often rescanned, merged and annotated along the way, so the weight of each test is adjustable by document type, and the score is a signal for the claims handler, never a verdict on its own. We do not publish a US or UK figure of our own on death certificates.
Source: ALFA, 2025 activity report published in July 2026; MoneyVox, top 3 frauds by line according to ALFA, July 2026; Koncile data 2026, all documents
A filled in template or a replaced date often shows no visible error. The checker relies on signals inside the file and the image, then on the consistency of the certificate with the rest of the claim.
| Element of the certificate | What Koncile checks | What gives a fake away |
|---|---|---|
| Issuing software | Creator and Producer, compared with how a vital records office or a scanner produces a copy | A form builder, a template site, a word processor or a consumer PDF editor |
| File dates | Creation and modification dates, compared with the date of issue | A file created or edited after the copy was issued, or rewritten dates |
| File writing style | Object order, cross reference table, stream compression, saved versions | The fingerprint of a known generator or editor, an earlier version with another name or date |
| File nature | Native PDF, scan, photo or hybrid file | A native PDF where a scanned certified copy is expected, an image pasted into a PDF |
| Image | On a scan or a photo: edited zones and AI generated parts, with K-Detect | A name or a date laid over the original, a seal or a signature pasted in |
| Certificate content | File number present, certifier named, dates in order | A date of death after the date filed, no file number, no certifier |
| Claim consistency | Name, date of birth and dates compared with the policy and the claim form | A name or a date that changes from one document to the next |
| External sources | Not queried by Koncile: the claims handler confirms with the issuing office | A record the office cannot find |
A death certificate holds a name, a date of birth, family details and often a cause of death. Founded by a former lawyer, Koncile was built with data protection at the core of its architecture: third party SOC 2 audit, GDPR compliance, secure hosting. Certificates are encrypted, deleted after analysis and never used to train a model.
Check the copy: a certified copy from the vital records office of the place of death, with a seal, a registrar's signature and a state file number. Check that names and dates match the claim. Then check the file: the software that produced the PDF, its dates, editing traces. Koncile runs the file checks in about ten seconds; the issuing office confirms the record.
For most claims, insurers and banks ask for a certified copy issued by the vital records office, not a photocopy. Many accept a scan or a PDF at first, then ask for the certified copy for large amounts. A scan is exactly where a file check helps: it tells you whether the copy you received was edited.
It can produce an image that looks like one, or change a name or a date on a real copy. What it does not produce is a record filed with a vital records office. Koncile looks for AI generated areas and edited zones on images, and for generator and editor fingerprints in PDF files.
Yes, when the fake death is used to obtain money. Insurance fraud is a crime in every state, and claiming a benefit through the mail or by wire can bring federal charges under 18 U.S.C. § 1341 or § 1343, punishable by up to 20 years in prison. In the UK, fraud by false representation carries up to 10 years under the Fraud Act 2006.
In the UK, the doctor issues the medical certificate of cause of death, which is used to register the death; the death certificate is then a certified copy of the register entry, issued by the registrar. In the US, the certifier's medical part sits on the death certificate itself, and the certified copy comes from the vital records office.
Often, yes. On a scan or a photo, the original PDF metadata is lost, but image analysis still finds edited zones, text laid over the original and AI generated parts. The content checks, dates, names and file number, run the same way.
It does not query registries, so it cannot confirm that a death was registered. It cannot see an authentic certificate used under a stolen identity. Foreign certificates get lighter content checks. A high score is a signal to verify with the issuing office, never a verdict.
Yes. Koncile complies with SOC 2 and GDPR, and an independent firm certifies our security and data protection. Your documents are hosted securely with a provider that itself holds SOC 2 and ISO 27001 certifications, and are deleted after analysis.
Each document gets its own method: a proprietary ML model for photos, known metadata and forensics for native PDFs, and contextual cross checks.
Inflated balances, deleted lines, a PDF rewritten with software other than the bank's.
See the detection
Net pay changed, contributions that do not match gross pay, a layout that does not match the payroll software.
See the detection
Photo replaced, date of birth or number changed, an MRZ that no longer matches the fields.
See the detection
Edited amounts, wrong VAT, duplicates and made up suppliers.
See the detectionUpload your bank statements, pay stubs or IDs: Koncile extracts the data and flags fakes in seconds.
Talk to an expert
