How to Spot a Forged Peptide COA

ON THIS PAGE

1. Key findings
2. What the Paradigm record establishes, and what it does not
3. Check one: the laboratory has to be a laboratory
4. Check two: the report number has to come back
5. Check three: the dates have to be true at once
6. Check four: one figure across a catalog is a claim about a business
7. What a clean check does not settle
8. Sources

Seventy months in federal prison for the owner, sixteen for his sister and employee. A $5 million money judgment against him alone, $78,317.52 in restitution owed by both, and more than 54,000 customers across all 50 states and more than 80 countries. That is the Paradigm Peptides sentencing, announced by the US Attorney's Office for the Northern District of Indiana on July 30, 2026, and it is why a forged certificate of analysis is not hypothetical here.

The checks that catch one are dull and quick. Does the named laboratory exist as a laboratory. Does the report number come back. Can all the dates on the page be true at once. Does the figure describe a lot or a catalog. None needs an instrument.

What the Paradigm record establishes, and what it does not

The record here is unusually complete, which is what makes it useful. The Justice Department release, read 2026-09-12, states that Matthew J. Kawa, 48, and his sister Jennifer L. Stechkober, 32, were sentenced by US District Judge Cristal C. Brisco after pleading guilty to introducing unapproved new drugs into interstate commerce with intent to defraud and mislead. The figures divide. Kawa drew the 70 months and the $5 million money judgment, and he also pleaded guilty to illegally importing merchandise; Stechkober drew 16 months. The restitution is joint and several, and the same office's case page dates both pleas to December 10, 2025.

On the documentation it is specific. On the business website and in emails to customers, it says, the operators falsely stated that the business was licensed and registered with the FDA, that it manufactured its products at its own laboratories in the United States, and that it tested them for quality. In fact the products were imported from Asia, including China and India, and were not tested before sale. Many items sold as selective androgen receptor modulators were later determined to contain testosterone. FDA warning letters went out in 2020 and 2022 and selling continued. One is published: the agency's letter to Paradigm RE LLC, dated December 7, 2020, names paradigmpeptides.com. That is the business meant here and nowhere else, and its site went offline in March 2024. CBS reports that similarly named storefronts have appeared since, and quotes one of them denying any affiliation.

The word forged is not in that release. It comes from CBS News, which covered the sentencing and wrote that Kawa admitted the business forged laboratory certificates. CBS does not say where that admission was made, and it names the plea deal explicitly when it means the plea deal: the same account quotes the deal conceding he knew the products were "marketed and sold for human consumption," which is the research-use disclaimer failing too. Keep the two apart: the sentence and the pleas are the government's, the forgery admission a news account.

Check one: the laboratory has to be a laboratory

A certificate naming no laboratory is unverifiable by construction, and that is a finding rather than a gap. A printed name is a claim about a third party, so it gets checked at the third party: an accreditation body's public register, or the laboratory's own site reached by searching its name. Where the mark names an accrediting body and a certificate number, those two strings are what you carry to that body. Never a link on the certificate, because a forged certificate can point at a forged portal. Paradigm is the instructive version, because the laboratory it claimed was its own. An in-house laboratory is not disqualifying, but nobody independent stands behind its document, and the seller's honesty is the one thing a certificate exists to remove from the question.

Check two: the report number has to come back

Some independent testing operations here publish part of their work, and the lookup is the check. Janoshik Analytical's public portal, opened 2026-09-12, is a listing of published tests, searchable through a single box, under a standing line saying the laboratory does not endorse, sponsor or have any affiliation with the companies, organizations or products listed. Read that line first: a hit proves a report with that number exists and says what the laboratory's copy says, and nothing about the seller.

Two strings get confused here, so keep them apart. The number the portal prints beside a listed test is the laboratory's task number, the same string printed on the vendor's copy, and it is what the search box matches: run against the listing on 2026-09-12, three of the 34 task numbers recorded in this publication's ledger returned their test, one of them carrying the client, manufacturer and sample the ledger records for it, and the other 31 returned nothing. The second string is a twelve-character verification key. The portal's entries link outward to an address built from the task number and that key, and the box does not index the key: it matches the printed number and the sample, client and manufacturer text, so typing a key into it is not a check at all. A miss on a task number is the ordinary outcome, not a finding: it means that test is not in the public listing. A hit is worth the comparison, and the comparison is the verification: compound, lot code, analysis date, method and every figure, read against the vendor's copy.

The identifier is where this thins out. Of the 126 supplier records in this publication's ledger, read 2026-09-12, 33 name Janoshik somewhere in their notes, and in 15 of those 33 the notes also carry an identifier string, counting the words verification key, verification code and task number together with the form the laboratory prints, task followed by a number. One of the 15 uses those words only to record a certificate that carries none of them. That measures the ledger's notes, not the market. Naming a laboratory is cheap. Printing the string that lets a stranger pull the laboratory's own copy is not.

Check three: the dates have to be true at once

A forger edits the number that sells the product. The fields nobody reads are the ones that cross-check each other, so that edit has to survive them. NIST's Weights and Measures crosswalk for ISO/IEC 17025 clause 7.8, dated February 2018 and read 2026-09-12, tabulates what a laboratory report carries unless the laboratory records a valid reason not to: unique identification with the end of the document clearly marked, the method, unambiguous identification of the item, the date of receipt, the sampling date where that is critical to the validity of the work, the date of issue, and who authorized it.

Those fields cross-check each other. An issue date earlier than the date of receipt is an impossibility on its face. The Janoshik reports described in this publication's ledger print three more: a date the testing was ordered, a date the sample was received and a date the analysis was conducted. An analysis date later than the day the lot went on sale describes testing that had not happened when the claim was made, and a lot code postdating the certificate that names it cannot have been the sample. An accreditation mark carries a date check of its own, and FDA states the shape of it for a different market: under its Accreditation Scheme for Conformity Assessment, which covers medical devices and radiation-emitting products rather than this one, a laboratory may label testing as conducted under the programme only where the FDA-recognized consensus standards and test methods used were within its scope of accreditation at the time of testing. Reading it across to a peptide certificate is this publication's inference and not FDA's, but the arithmetic holds: expired on the day of the run is a failure.

Check four: one figure across a catalog is a claim about a business

A purity figure belongs to one lot, by one method, on one day. When the same percentage sits on every product page in a catalog, it has stopped describing material. CBS reported that before the site went down in March 2024, Paradigm advertised its offerings as made in America, 99 percent pure with no analytical method named anywhere in that description, and third-party tested. A purity figure with no method beside it is uncheckable.

The pattern is not rare. Some form of the phrase catalog-wide appears in 37 of the 126 supplier records in this ledger, read 2026-09-12, and 33 of those records use it for a claim asserted across a catalog rather than tied to a lot. The other four use it only for the opposite finding, or for pricing. One question closes it: which lot is this document about, and is that code on the vial.

What a clean check does not settle

A verified report proves a laboratory analyzed a sample and reported those figures. It does not prove the sample came from the lot that ships, or that other vials match the one tested. Those are sampling questions and no portal answers them.

The ledger counts describe 126 supplier records read on 2026-09-12 and state what those notes contain, not what the market does; a supplier absent from the ledger is unmeasured rather than cleared. The forgery admission is CBS's reporting of what Kawa conceded, and the Justice Department release does not use the word. ISO/IEC 17025:2017 was not read directly here: its clause 7.8 content is reported as NIST's crosswalk tabulates it.

Three of the four checks above were open to Paradigm's customers across the whole April 2019 to March 2024 window the prosecutors' case page gives for purchases. The laboratory check would have returned nothing, because the laboratory it named was its own, and that absence was the finding. The next case turns on whether anyone runs them.

Sources

All read 2026-09-12. US Attorney's Office, Northern District of Indiana: the press release of July 30, 2026 announcing the Paradigm Peptides sentences, for the terms, the money judgment, the restitution, the customer counts and the documentation findings, and the same office's case page for United States v. Matthew Kawa, for the December 10, 2025 pleas and the April 2019 to March 2024 purchase window. US Food and Drug Administration: warning letter 612014 to Paradigm RE LLC, dated December 7, 2020, and "Testing Laboratories: How to Participate in ASCA," for the scope-at-time-of-testing rule and for that programme's stated coverage of medical devices and radiation-emitting products. CBS News, "Judge sentences peptide vendor to nearly 6 years in prison for 'an incredible trail of harm,'" updated July 30, 2026, for the forged-certificate admission, the quoted plea-deal line, the March 2024 shutdown, the advertising claims and the successor site's denial of affiliation. National Institute of Standards and Technology, Weights and Measures crosswalk for ISO/IEC 17025 clause 7.8, February 2018, for the reporting elements listed above. Janoshik Analytical's public report portal, for its standing non-affiliation line, the fields its search box matches, and the ledger task numbers run against its listing. Supplier counts come from this publication's own vendor ledger of 126 supplier records. No certificate was examined and no vendor's site was read for this article; the only laboratory site opened was that laboratory's public portal, and ISO/IEC 17025:2017 itself was not read.