The 2026 ad rules across TikTok, Reels, and Shorts

• —AI voice cloning must be labeled. If your voiceover uses AI to replicate an identifiable person's voice, the ad must carry TikTok's AIGC label or a clearly visible caption indicating the audio is AI-generated. Unlabeled AI voice clones result in ad rejection.
• —Public-figure likeness requires permission. You cannot use the likeness, voice, or image of any public or private figure in an ad without explicit permission — including AI-generated depictions. This is enforced more aggressively on ads than on organic content.
• —Organic vs paid policy gap. A skincare transformation video might be compliant organically but violate ad creative policy when boosted as a Spark Ad (before-and-after imagery rules are stricter for paid). Run the ad-creative-policy check before pressing boost.
• —Spark Ads — flexible technical spec. No restrictions on ratio, resolution, duration, file type, or file size. Sparked videos can be up to 10 minutes. Each TikTok Ads Manager account supports up to 10,000 Spark Ads. Captions can be blank; account tagging, emoji, and hashtags allowed in the caption.
• —In-Feed standard spec. Vertical 9:16 (1080×1920 px), 5-60s recommended (up to 60s ideal), MP4/MOV/MPEG/AVI/GIF, ≤500MB.

• —9:16 vertical, ≤90s, 4GB file cap. Recommended 1080×1920 px. MP4, MOV, GIF, or static JPG/PNG for image-based variants. (Stories shares the 9:16 canvas but caps at 120s; standard Feed allows up to 240 minutes.)
• —Unified Stories/Reels safe zone (March 2026). On a 1440×2560 canvas: keep top 14% (~358 px), bottom 20-35% (~512-896 px), and 6% on each side (~87 px) free of text, logos, and key creative elements. The profile icon + CTA bar cover the bottom; the action stack lives in the side margins.
• —Organic music library is off-limits for ads. Tracks from Instagram's consumer music library can't be used on paid placements. Use the commercial-cleared Sound Collection inside Meta Ads Manager or upload your own licensed track. User-generated original sounds require explicit written permission from the original creator.
• —Sound-on is more dominant than on TikTok. Meta's own data shows sound-on Reels ads outperform silent equivalents by ~35% in engagement. Voiceover-first Reels outperform music-heavy Reels by ~23% in DTC conversion. Calibrate the audio investment accordingly.

• —Higher brand-safety bar than long-form. Content that's monetizable on long-form YouTube can still be demonetized as a Short. The Shorts feed environment has stricter advertiser-friendly thresholds — re-check compliance even if your long-form videos pass.
• —Non-original Shorts are demonetized. Unedited clips from TV/movies, reuploads of other creators' content, and views from automated bots are all ineligible for the Shorts ad-revenue share. Add commentary, editing, or transformative value to make a Short eligible.
• —Don't fake organic engagement. Shorts ads must not mimic organic content in ways that obscure the commercial nature. Using fake comment overlays or fake like counts on the creative to simulate organic engagement is classified as deceptive ad behavior — a fast path to rejection.
• —VAC is now Demand Gen. Google renamed Video Action Campaigns to Demand Gen. The campaign type still covers Shorts but now also expands into Discover and Gmail placements. Reach is broader, so creative should hold up outside the Shorts shelf too.
• —Spec basics. Vertical 9:16, ≤60 seconds, audio must be either YouTube Audio Library / Creator Music or properly licensed original audio. Title appears below the thumbnail in the Shorts shelf — write it as a hook, not as a topic label.

• —Disclose every material connection. Cash payment, flat fees, ongoing sponsorships, free products, affiliate commissions, perks (trips, tickets), business relationships, and personal relationships tied to the promotion all count. If a reasonable consumer would want to know it before relying on the recommendation, disclose it.
• —"Clear and conspicuous" is enforced. A #ad buried at the end of a long hashtag list is NOT compliant. Place the disclosure where the viewer sees it before engaging — ideally at the start of the caption, the first 3 seconds of the voiceover, or as on-screen text on the opening frame.
• —All creators, all follower counts. FTC rules apply equally to nano-creators (1K followers) and mega-creators (10M+). The agency has been actively enforcing against small accounts — influencer-related FTC cases were up ~340% in 2025 vs 2021, with over 150 warning letters issued in 2025 alone.
• —AI-generated endorsements need their own disclosure. Per FTC guidance from late 2024 (refined in 2026): if a creator or brand uses AI to generate or significantly alter endorsement content (voice, likeness, performance), that AI involvement must itself be disclosed. Disclosing "ad" alone is not enough when AI was used.
• —Penalty ranges. FTC penalties for inadequate disclosure run $5,000 to ~$43,792 per violation depending on severity and reach. The brand AND the creator can both be liable.

The Ad Bench rubric flags missing or weak disclosure under risk flags → before_after_disclaimer and risk flags → deceptive_pattern; affiliate-flagged ads see disclosure weighted heavier into the CTA + brand-fit scores as well.

A 9-point pre-launch check that catches the most-flagged issues across all three platforms:

• 1.Aspect ratio. 9:16 vertical, 1080×1920 native. Non-vertical creative gets letterboxed or rejected.
• 2.Duration. Plan for the strictest cap (Shorts at 60s). Reels accepts 90s; TikTok Spark Ads accept up to 10 minutes but in-feed convention is still ≤60s.
• 3.Audio licensing. Use commercial-cleared tracks (TikTok Commercial Music Library, Meta Sound Collection, YouTube Audio Library) or upload your own. Don't paste a trending UGC sound into a paid placement.
• 4.FTC disclosure placement. On-screen text or voiceover in the first 3 seconds. Repeated in the caption. Not buried in a hashtag stack.
• 5.AI-content labels. If voice, face, or core performance was AI-generated, use the platform's AIGC label AND visible on-screen text saying so.
• 6.No fake engagement overlays. Don't composite fake comment counts, fake like icons, or fake share/save UI into the creative. Shorts and TikTok both flag this as deceptive.
• 7.No public-figure likeness without permission. Including AI deepfakes. Including dead public figures. Including news anchors / commentators reading a fake script.
• 8.Before/after imagery hygiene. Disclaimers visible ("results not typical"), treatment timeframe stated, no implied medical claims. TikTok and Meta both reject ads with unsupported transformation imagery.
• 9.Sensitive category review. Health claims, financial-services promises, alcohol, supplements, prescription-tier products all require extra eligibility. Some are banned outright on Shorts; others need geo-gating on Reels.

This article aggregates current public policy + creative-spec documentation. For the authoritative text, link directly:

• —TikTok — About Spark Ads (official Help Center).
• —TikTok — Auction In-Feed Ads (Ads Manager spec).
• —Meta — Instagram Reels Ads Guide (specs, safe zones).
• —YouTube — Advertiser-friendly content guidelines (monetization eligibility).
• —YouTube — Shorts monetization policies.
• —FTC — Endorsements, Influencers, and Reviews (regulator's own page; this is the source of truth).

Rules change. The Ad Bench rubric calibration tracks platform policy updates — but you should still hit the official docs above before launching a sensitive-category campaign. When in doubt, consult counsel; The Ad Bench is a creative-review tool, not a legal-advice service.