How We Actually Verify a Claim
This is the page that is, honestly, the whole brand: not "we're careful," but the specific, repeatable pipeline every issue goes through before it publishes — including how AI is and isn't involved. See also editorial standards (the 10 principles) and corrections policy (what happens when we get it wrong anyway).
The verification pipeline
Every issue of The Peptide Current — including Issue No. 01 — goes through the same five steps before it publishes:
-
01
AI drafts using primary-source web search. We use AI tools to search FDA databases, the Federal Register, ClinicalTrials.gov, PubMed, SEC filings, court dockets, and peer-reviewed journals, and to draft an initial version of the piece grounded in what those sources actually say.
-
02
Every factual claim gets logged. Every claim in the draft — not a sample, every one — is entered into a verification log: the claim, the primary source it rests on, and the URL. This is modeled directly on the log we published for Issue No. 01; see the example.
-
03
Claims that can't be verified get cut, not softened. If a claim looked good in the first draft but the primary source doesn't actually support it — wrong number, wrong study population, a source that turns out to be secondary coverage rather than primary — it comes out. Those cut claims are logged too, with the reason, in the same verification log. That "claims considered but cut" section is itself part of the discipline: it's the clearest evidence that the verification step is real and not decorative.
-
04
A separate system re-verifies every factual claim before publication. Nothing goes straight from the drafting pass to the public site. A second AI system checks the draft and verification record against the cited primary sources, and publication remains a separate manual, non-automatic action.
-
05
The verification log publishes alongside the issue — public, not on request. Every issue gets its own verification log page, linked from the article itself. This was the single biggest gap in earlier drafts of this site (the disclaimer used to say the log was "available on request" — it isn't anymore, it's a real page). A publication that says "trust our sourcing" without showing the sourcing is just asking for trust it hasn't earned. This is how we earn it instead.
What this means in practice
AI systems support both research and drafting and the separate verification pass. One system prepares the source-grounded draft; a second system independently rechecks its factual claims against the primary record. Publication is manual and never automatic. PepCurrent does not describe that second-system check as human review. If a claim can't be traced to a primary source, it doesn't run, no matter how well-established it "feels." See our AI disclosure for the plain-language version of this same commitment.
Legal and medical scope
The Peptide Current is educational journalism, not medical advice, and not a place to buy anything. A few rules that follow from that, in plain language (the fuller legal-register versions live in Disclosures and Editorial Standards):
- We report what a study did — "researchers used X over Y weeks" — as journalism about the study, never as advice about what a reader should do. Every piece carries a visible reminder that this is not medical advice and we don't recommend for or against use of any substance.
- We do not sell peptides, supplements, or anything else, and we never link to a vendor, compounding pharmacy, or gray-market seller. Where a vendor is named because they're the subject of an FDA enforcement action, any link goes to the FDA's own page about that action — never to the vendor.
- We currently have zero affiliate or sponsor relationships. If that changes, FTC clear-and-conspicuous disclosure applies: the relationship gets disclosed right next to the claim, not tucked into a separate page.
Comments — a future-state rule, set now
We don't have comments or any other user-generated content today. If we ever add them, the policy is decided in advance, not improvised after something goes wrong: pre-moderation on every comment, a clear conduct policy, and no medical claims allowed from commenters — full stop. Writing this down now, before the feature exists, is the point.