04

drip · content safety

content safety policy.

effective · 2026-04-18 · version 0.1

what this policy is

drip's content safety layer is pre-publication, automated, and hard-enforced. it runs before brand generation, after generation, before publish, and nightly against live content.

agents cannot override it. creators can appeal individual decisions.

this policy applies to every surface drip publishes on behalf of a creator: storefronts, product pages, ad creative, marketing copy, and agent outputs.

hard blocks

therapeutic claims — any claim that a product diagnoses, treats, cures, or prevents a disease. fda drug jurisdiction triggers on therapeutic claims. cosmetic tier ends, regulatory pathway becomes drug approval.

unsupported efficacy claims — any numeric magnitude without supporting clinical data. "visibly smooths" or "supports the appearance of" is cosmetic and fine. "reduces wrinkles by 40 percent" without data is blocked.

minors — no products targeted at users under 18. no imagery of minors in product contexts. no claims addressing adolescent skin, puberty, or growth.

prescription drug comparison — "like ozempic without a prescription" or "natural alternative to viagra" drags the cosmetic tier into drug territory. blocked.

restricted substances — controlled substances as ingredients, anabolic steroid adjacent framing, or growth-hormone therapy framing on ghrp-style peptides.

identity targeting — claims that tie product efficacy to race, ethnicity, or national origin. exclusionary marketing is blocked.

soft blocks

adjacent-medical framing like "supports hair growth" or "collagen production support" is routed to a human-review queue. most pass. "immune-boosting" usually does not.

influencer-style puffery — "life-changing," "miracle," "guaranteed results" — might be fine on social but triggers regulation-by-enforcement at scale. reviewed case by case.

celebrity or expert endorsement requires proof. "as recommended by dr. name" or "used by athlete during training" or "featured in publication" must be supported before publish.

warn and publish

before-and-after imagery is allowed but flagged for an authenticity check.

subscription auto-renew language must be disclosed. drip auto-adds the required disclosure.

ingredient-novelty claims like "world's first" are allowed if true.

tier-specific rules

cosmetic tier — every rule in §02 and §03 applies. product categories limited to topicals, skincare, haircare, body care.

research-chemical tier — mandatory disclaimer "for research use only. not for human consumption." on every product page and ad. no human-dosing recommendations. no body-modification outcomes. msds availability required. 18+ age gate at checkout.

how enforcement works

pre-generation — the brief itself is checked. briefs containing therapeutic language are rejected before a model runs.

at generation — every ai output runs through the safety layer before returning to the creator. blocked outputs show a [regenerating] state and re-sample automatically.

at publish — the final publish-gate check runs before a storefront goes live or ads reach meta, tiktok, or youtube.

post-publish — drip scans live content nightly. if rules change, the creator gets 48 hours to revise or the platform takes the content down.

appeals

creators may appeal any safety decision to safety@drip.markets. target response is 48 hours. escalation is to a human reviewer, never an agent.

an upheld appeal moves content from blocked to published. repeat appeals on the same claim type within 90 days route to platform-level review.

why this matters

drip's legal position — that the platform is rails, not seller-of-record — depends on the content safety layer working. if the platform knowingly publishes therapeutic claims, the rails-only defense weakens.

this is the compliance moat that lets drip operate at creator speed without fda enforcement risk landing on the platform. every creator's interest is aligned: a lawsuit against one creator that names drip as co-defendant threatens every other creator's brand.

transparency

drip publishes quarterly stats: briefs rejected pre-generation, outputs regenerated mid-flow, appeals filed and appeals upheld, post-publish takedowns.

no creator is named in public stats. internal rate-limiting watches for systematic abuse.

effective · 2026-04-18. contact safety@drip.markets.