Before-and-After Product Demo Ads: Is AI Allowed on Meta and TikTok?

None of the pages read on 2026-10-03 bans a before-and-after ad because AI made part of it. They look at the product category, the claim and whether the picture is true. So the practical answer for a product demo is: film the real before and the real after, and let AI build everything around them.

Sources: Meta's Health and Wellness standard, TikTok's healthcare ad policy and the FTC business blog Less than meets the eye? (January 23, 2014). This is a reading of public policy pages, not legal advice.

What each source says

Meta's page, whose changelog shows an edit on July 23, 2026, permits cosmetic products, procedures and surgeries depicting a before-and-after transformation when the ad targets people 18 or older. It also bars clickbait in a health context, including promises of specific outcomes within a set timeframe without disclaimers or qualifiers.

• TikTok's page is organized by market. Look up the country you target before you assume a rule applies or does not.

Why a generated after-shot is the risk

If the clip claims to show what the product did, the FTC's test is whether the visual is a truthful representation of what it can do. A generated stain that vanishes in one swipe is a depiction you made up, not a demonstration. A generated kitchen around a real wipe is set dressing. Film the surface twice with the same camera and light, keep the result unretouched, and use AI for the hook, the room and the voice.

Label what is staged, and keep it on screen

The FTC's Nissan example shows how weak a fleeting disclaimer is: the line appeared for three seconds. Sume's caption job can burn authored cues with start and end times, so a label can stay up for the whole shot. Standalone captions are $0.20 for a video up to 60 seconds, and the clip must be a public HTTPS URL. Replace the wording below with a statement that is true for your footage.

curl -X POST https://api.sume.com/v1/video-captions \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: demo-label-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/example/demo.mp4",
    "style": "slam",
    "cues": [
      {"text": "REAL FOOTAGE, ONE WIPE", "start": 0.0, "end": 6.0},
      {"text": "RESULTS VARY BY SURFACE", "start": 6.0, "end": 12.0}
    ]
  }'

For the build itself, see the before-and-after video API and one real action in a usage demo. For platform labels on the AI parts, use the disclosure checklist.