Compliance centre

FDA has taken enforcement action against sellers where “research use only” and “not intended for human consumption” labelling was contradicted by other evidence of intended human use on the same website. The FTC separately evaluates the overall net impression of health-related marketing, including graphics, testimonials, charts, and implied claims, and has stated that qualifying language cannot cure a contradictory overall impression.

The conclusion we drew is that a disclaimer is not a compliance strategy. It is a label on top of whatever the product actually does. So the constraints are built into the system:

• Content risk level, publication status, and compliance status are database fields, not editorial conventions.
• No page states a recommended quantity. Quantities appear only inside study records, attached to the study that reported them.
• Calculators present solution-preparation arithmetic over values the user supplies, and never generate a suggested quantity.
• The schedule tool converts a schedule the user already has into dates. It has no opinion about what the schedule should be.
• Personalisation matches users to study populations. It cannot output an individualised recommendation because no code path produces one.
• Feature flags allow any category to be disabled globally, immediately removing it from pages, search, and recommendations.

The following are treated as high-risk and are excluded from platform copy. Where they appear in this catalogue at all, it is inside a direct description of what a source reported, never as our own assertion.

The word safe deserves specific mention. We do not describe any compound as safe, and we never convert “no documented interaction” into “safe together”. The register renders that state as: No interaction meeting our evidence criteria is currently documented. Absence of evidence is not evidence of safety.

• Compensation never influences an evidence grade, a claim, a study selection, or a ranking. Rankings are computed from research relevance, evidence confidence, and content quality.
• Editorial, utility, and commercial recommendations are visually distinguished from one another.
• Disclosure appears adjacent to the recommendation itself, not behind a link or in a footer.
• We do not sell peptides, medications, or research compounds, and we do not take payment to change a rating.

• Preferred sources: PubMed-indexed primary literature, systematic reviews, meta-analyses, ClinicalTrials.gov, FDA, FDA-approved labelling, DailyMed, openFDA, and RxNorm.
• Human and non-human evidence are never pooled. Routes, formulations, endpoints, and materially different populations are kept separate.
• Where a source identifier has not been verified, we publish no identifier rather than a plausible-looking one. See the study register.
• Unverified records are excluded from every derived statistic.

The current build has no accounts and no server-side storage. The research profile, theme preference, and age acknowledgment live in your browser’s local storage and are never transmitted. We do not have a copy of your profile, because there is nowhere for it to go.

When accounts are introduced, research-profile data will be treated as the most sensitive class in the system: row-level security tied to the owning user, no advertising or affiliate system ever receiving it, export and deletion controls, and privileged-access logging for any support access.