Compliance basics — FDA, FTC, platforms, and why "legal" isn't a single thing

The plain-English version

When marketers talk about "compliance," they mean "things that can get us in trouble." For a GLP-1 telehealth brand, "trouble" comes from three different places, each with different consequences:

1. FDA (Food and Drug Administration) — controls what you can say about a prescription drug. Penalty: warning letter, public shaming, and the threat of legal action ("seizure and injunction" per FDA's Feb 2026 communication).
2. FTC (Federal Trade Commission) — controls whether ads are deceptive. Cares about substantiation, comparative claims, testimonials, and earnings claims. Penalty: consent decrees, fines.
3. Platform policies — Meta, Google, TikTok all have their own rules. Penalty: ad rejected → account flagged → account suspended. Often stricter than the law.

Pass the law but fail the platform = your ad doesn't run. Pass the platform but fail the law = you get a warning letter and possibly sued. You have to clear all three.

FDA — what they care about

The FDA regulates prescription drug promotion through the Office of Prescription Drug Promotion (OPDP) and the Office of Compounding Quality and Compliance. For our vertical, the cared-about issues are:

• Sameness-with-branded claims. Saying or implying that compounded GLP-1 is "the same as" Wegovy/Ozempic/Mounjaro/Zepbound. This is the single most-cited violation in 2025-2026 enforcement waves.
• Implied FDA approval. Calling a compounded product "FDA-approved" or describing it as "clinically proven." Compounded preparations are not FDA-approved, full stop.
• Off-label promotion. Promoting GLP-1 for non-FDA-approved uses (longevity, mood, addiction). Don't do this in copy.
• Fair balance. Mentioning the medication without mentioning side effects is treated as deceptive.

The pipeline has 25 GLP-1-specific rules in data/kb/regulatory/glp1-guardrails.md. The compliance node also references 8 curated FDA warning letters in data/kb/regulatory/fda/ so the judge sees what the FDA actually penalized.

FTC — what they care about

The FTC focuses on deception: would a reasonable consumer be misled by this ad? Their concerns overlap with FDA's but extend to:

• Substantiation. Health claims need scientific backing. "Lose 35 lbs in 12 weeks" needs evidence specific to the compounded product (which generally doesn't exist).
• Testimonials. Per the FTC Endorsement Guides (16 C.F.R. Part 255), testimonials must reflect typical results or disclose what is typical. Before/after photos used as headlines are a systematic FTC concern.
• Comparative claims. "Affordable alternative to Wegovy" is a comparative claim that creates implied therapeutic substitutability (SimpleRx warning letter, Sept 2025).

State versions of the FTC Act exist too — California Business & Professions Code §17508 specifically prohibits unsubstantiated weight-loss claims.

Platforms — stricter than the law

This is the part that surprises most non-marketers. Meta, Google, and TikTok all have their own policies, and they often go BEYOND what the FDA or FTC requires. Why? Because platforms face their own pressure (state AGs, regulatory scrutiny, PR risk) and they err on the side of restriction.

Examples of platform-only rules:

• Meta requires LegitScript certification AND its own internal pre-approval, for the prescription GLP-1 category. Neither is legally required by FDA — they're Meta's gates.
• Google prohibits bidding on competitor brand terms (Wegovy, Ozempic) for compounded advertisers. Not a legal requirement; a Google policy.
• TikTok bans prescription weight-loss advertising outright. Not a legal requirement; a platform decision.

You can be fully FDA/FTC compliant and still get your ad rejected. You can also get rejected for things that wouldn't be a federal-law issue at all.

The three-stage check the pipeline does

The compliance node in this pipeline implements a layered approach:

1. Regex screen — deterministic, free, catches the obvious. About 12 rules have reliable regex patterns ("same active ingredient as Wegovy" matches a specific phrase pattern).
2. Semantic judge (LLM) — for the 14 rules where meaning matters more than wording. Sees the FDA enforcement precedents and the platform policy for the target channel.
3. Persona-specific threshold — Rachel gets 0.95, Maya/David get 0.80, Jess gets 0.85. The same finding can pass or fail depending on who the copy is for.

The output is a pass / warn / block decision, with concrete suggested_fix language the copy_polish node can act on.

What to remember

Three rules of thumb that will save you most of the trouble:

1. Don't say "same as" a branded drug. Ever. In any channel.
2. Don't promise weight numbers ("lose X lbs in Y weeks"). The substantiation needed doesn't exist for compounded products.
3. Don't frame the clinician interaction as a friction ("skip the doctor visit," "get a prescription in minutes"). This undermines the legal basis for compounded prescribing AND violates platform speed-of-access policies.

If you remember nothing else from this lesson, remember those three.

Primary sources — go straight to the regulators

The pipeline summarizes these in data/kb/regulatory/fda/ and data/kb/regulatory/platforms/, but the originals are short and specific. Read the source at least once.

• 🏛 FDA Office of Prescription Drug Promotion (OPDP) — the office that issues warning letters about prescription-drug ads.
• 🏛 FDA — Concerns with Unapproved GLP-1 Drugs — FDA's standing guidance on our vertical.
• 🏛 FDA Warning Letters search — filter by OPDP to find the GLP-1 letters. The actual documents.
• 🏛 FTC Endorsement Guides — the testimonial / influencer / before-after rules.
• 🏛 FTC Health Products Compliance Guidance — the substantiation rules in plain English.
• ✅ LegitScript Healthcare Merchant Certification — the certification Meta and Google both require for prescription GLP-1 advertising. You can't run paid ads without it.

For platform-side policy documents (Meta, Google, TikTok, Microsoft), see Resources.

What to do next

• Browse data/kb/regulatory/fda/ to see the real warning letters the judge references.
• Read GLP-1 specific gotchas for the vertical-specific traps.
• Then Brand voice for how Eudaven's voice was designed in part to navigate these rules.