---
title: AI-generated ad creative — where Barry Hott uses it and where the lines are
slug: ai-generated-ad-creative
summary: How Barry Hott used Veo 3 / Kling in 2025 — AI hooks followed by real-human ads, B-roll and stock-footage replacement, animating existing product stills, AI "pre-shoots" before expensive shoots, prompting exact pacing and inflection — the compliance line (no AI testimonials about physical experience), why the bar for "pretty ads" dropped to ~zero, his worry about AI static-image glut, and Savannah Sanchez's no-AI counterpoint.
domain: marketing
tags: [ad-creative, ai, veo, ugc, compliance, production-cost]
related: [ai-tools-for-app-ad-creatives, ugly-ads-and-brand-caution, creative-hit-rate-and-spend-tiers, hook-anatomy, ugc-ad-formula-and-formats, applovin-ad-creative]
sources: [yt:4BOtccKnC2Q, yt:I0SPikVQdYs]
updated: 2026-10-24
---

# AI-generated ad creative

## TL;DR
- Barry's main 2025 use: **AI-generated hooks followed by real human footage**. He ran a few fully AI ads, some at scale, but full end-to-end AI ads were not his goal.[^1][^2]
- **Compliance line:** no AI "people" describing a physical experience of the product, which is a fake testimonial (illegal the same way paying a creator to make false claims is). AI can tell a story or state a problem.[^3]
- Cheapest wins: **animate existing product/studio stills** (push-ins, product turns), generate B-roll you wish you had, and replace stock footage, which he says now has ~zero value to him.[^4][^5]
- Use AI to **pre-shoot**: prompt the exact pacing, inflection and setting of a hook ("say it faster", "more incredulous"), test it, then reshoot winners with a real creator.[^6][^7]
- Goal: lower cost per ad → more ads → more chances to win.[^8]

## Details
**Two kinds of AI content.** (1) AI creative that doesn't look AI-made. (2) Openly AI content people enjoy *because* it is AI (skydiving gorillas, cavemen). There is an audience for the second, but it may only last while it's novel.[^9] Barry's rule stays the same: make ads in the style of what consumers are consuming.[^10]

**Who notices.** He believes ~85% of people won't notice or care when it's done well. In his view the only audience reliably put off by realistic AI ads is people in film and creative work.[^11]

**Tools (mid-2025).** Google Veo (Veo 3 made "passable" talking UGC clips days after launch) and Kling. He recommends every brand at least play with them to see what's possible.[^12][^5]

**Pretty ads get cheap.** A shoot an agency would price at $1K–$50K can now cost "effectively 10 cents". That threatens high-end brands and production agencies. Barry still doesn't expect the broad ~80–85% of consumers to drift toward polished content: someone FaceTiming you from the street still beats an obvious sell.[^13][^14]

**The static-image concern.** AI makes image ads so easy that the supply of AI statics far outstrips how much image content people consume organically. He wonders whether audiences will tune out image ads altogether.[^15]

**Brand trust and performance marketing.** If you won't, a smaller competitor will. He has measured large brand lift from highly relevant performance ads in brand-lift tests, and argues volume-driven AI testing feeds brand rather than cheapening it.[^16][^17]

## Disagreements / caveats
- Savannah Sanchez's agency uses **no AI**, only its 40 in-house creators. When AppLovin pulled top performers for her brands, nearly all were strictly UGC-style.[^18] An AppLovin rep on the same call suggested AI video as a stop-gap to iterate on winning concepts.[^19]
- Barry flags the environmental cost (energy, water; Meta's 20-year nuclear deal) as an unsolved problem.[^20]
- This is an unedited mid-2025 conversation with Tina Donati. Tools and platform rules (e.g. YouTube demonetising AI content, Meta AI labels) change fast, so re-check.[^21]

## See also
- [[ai-tools-for-app-ad-creatives]] — app-side tools (ElevenLabs, Suno, Higgsfield) and FTC risk
- [[ugly-ads-and-brand-caution]] · [[creative-hit-rate-and-spend-tiers]]
- [[people/barry-hott]] · [[people/savannah-sanchez]]

[^1]: yt:4BOtccKnC2Q @ 03:58 — Barry Hott
[^2]: yt:4BOtccKnC2Q @ 33:33 — Barry Hott
[^3]: yt:4BOtccKnC2Q @ 04:49 — Barry Hott
[^4]: yt:4BOtccKnC2Q @ 17:37 — Barry Hott
[^5]: yt:4BOtccKnC2Q @ 33:33 — Barry Hott
[^6]: yt:4BOtccKnC2Q @ 22:23 — Barry Hott
[^7]: yt:4BOtccKnC2Q @ 24:30 — Barry Hott
[^8]: yt:4BOtccKnC2Q @ 19:13 — Barry Hott
[^9]: yt:4BOtccKnC2Q @ 05:39 — Barry Hott
[^10]: yt:4BOtccKnC2Q @ 07:17 — Barry Hott
[^11]: yt:4BOtccKnC2Q @ 29:33 — Barry Hott
[^12]: yt:4BOtccKnC2Q @ 01:53 — Barry Hott
[^13]: yt:4BOtccKnC2Q @ 25:36 — Barry Hott
[^14]: yt:4BOtccKnC2Q @ 27:11 — Barry Hott
[^15]: yt:4BOtccKnC2Q @ 31:55 — Barry Hott
[^16]: yt:4BOtccKnC2Q @ 16:02 — Barry Hott
[^17]: yt:4BOtccKnC2Q @ 20:46 — Barry Hott
[^18]: yt:I0SPikVQdYs @ 44:58 — Savannah Sanchez
[^19]: yt:I0SPikVQdYs @ 43:37 — AppLovin rep (Jonathan)
[^20]: yt:4BOtccKnC2Q @ 12:45 — Barry Hott
[^21]: yt:4BOtccKnC2Q @ 01:53 — Barry Hott