AI in Advertising: How Brands Use AI for Ads (with Examples)

Last checked: 29 September 2026

AI in advertising means using generative AI to create ad content such as video, images, voice and music, and using machine learning to decide who sees which version. Since 2024, brands from Coca-Cola to Deutsche Telekom have run AI-made campaigns, and an IAB survey from July 2025 found that 86% of video ad buyers use or plan to use generative AI for their video creative.

• Creative production: AI generates video shots, images and CGI-like scenes, which cuts production time and cost. Kalshi’s NBA Finals spot took two to three days.
• Voice and music: synthetic or reconstructed voices and AI-composed music. Deutsche Telekom used ElevenLabs to recreate Albert Einstein’s voice; Popeyes used Suno for a diss track.
• Versions and personalisation: one campaign becomes many variants for different audiences, formats and contexts. In the IAB survey, 42% of buyers create audience-specific versions.
• Virtual models and presenters: AI models or avatars replace a shoot, as in Mango’s Sunset Dream campaign or Austria’s AI-Sisi.
• Media buying and targeting: AI ranking and bidding systems decide who sees which ad, for example Meta’s Advantage+.
• Testing: teams generate hundreds of variants and keep the few that work. Kalshi’s creator made 300 to 400 generations to get 15 usable clips.

For the eight campaigns we rate highest, with the tools behind each spot, see our list of the best AI commercials of 2026.

• 86% of video ad buyers use or plan to use generative AI for video creative, and about half already do (IAB, July 2025). Buyers expect AI to build around 40% of ads by 2026.
• 98% of 201 German agencies use generative AI, 95% in creative work (BVDW, April 2025).
• Meta’s generative AI video ad tools reached a combined $10 billion annual revenue run rate in the fourth quarter of 2025 (Meta, January 2026).
• Global ad spend is forecast at $1.3 trillion for 2026, with AI named as the main disruptor (WARC, January 2026).
• Klarna reports that AI cut its image production cycle from six weeks to seven days (company figure, May 2024).

AI saves time, but audiences notice when it replaces craft without adding anything. Coca-Cola’s 2024 remake was criticised as soulless. McDonald’s Netherlands pulled its 2025 holiday ad within three days. Under Armour had to add credit for footage it had reused from an earlier shoot. Vodafone Germany’s AI influencer on TikTok drew attention for visible glitches and, according to Fortune, was not clearly labelled.

The lesson from these cases: use AI where it makes an idea possible, show the people behind it, and label what needs labelling.

Yes, in many cases. Article 50 of the EU AI Act has applied since 2 August 2026. Advertisers who publish a deepfake, meaning realistic AI-generated or manipulated images, audio or video of people, places or events, must disclose it clearly and at the latest at first exposure. Clearly artistic, satirical or fictional work may be labelled in a way that doesn’t spoil it. Fines can reach €15 million or 3% of worldwide turnover.

In Germany, a breach can also count as unfair competition, so competitors and associations can take action. The Wettbewerbszentrale published a labelling guide in 2026 and advises labelling when in doubt, in German for the German market. Tool providers have until 2 December 2026 to add machine-readable marking to AI systems launched before 2 August 2026 (Digital Omnibus on AI, Regulation (EU) 2026/1744). The advertiser’s duty to disclose deepfakes is not delayed.

Need the labels for your own AI ads? Download our free AI disclosure label kit for the EU AI Act.

1. Text on signs, packaging or screens is garbled or missing.
2. Very fast cuts or slow motion hide transitions, because AI clips are short.
3. Objects vanish, merge or break the laws of physics.
4. Faces, hands or small details such as moles change from shot to shot.
5. The platform shows a label: YouTube, for example, marks realistic synthetic content and can read C2PA metadata.

These signs are getting less reliable as the models improve, which is exactly why the EU now requires disclosure.

An AI commercial still starts with a brief, a script and a storyboard. The difference is in production: instead of a shoot, scenes are generated, selected and edited, and voice and music can be produced the same way. Our guide AI commercial production from idea to broadcast walks through each step, and our comparison of the best AI video generation tools shows which models are used for what.

Most scenes in a professional AI commercial are not generated from text alone. Image-to-video turns a product photo, a key visual or a generated frame into a moving shot, and video-to-video transforms real footage while keeping its movement, so the brand, the product and the people stay recognisable. We show how this works in Video2Video production with AI. Leading models for this in 2026 include Seedance 2.0, Kling 3.0, Google Veo 3.1 and Alibaba’s Happy Horse, and open models such as LTX can even run on your own hardware when footage must not leave the company.

Generating clips is the easy part. What separates a professional AI commercial from AI slop is consistency and craft: the same faces, products and light in every shot, a storyboard that holds the story together, and editing, sound design and colour grading that turn single clips into one film. That is where most of the work goes.

In our own work, AI is rarely the whole film. For Harley-Davidson Kiel, the dealership’s real staff and van appear in the commercial, while the delivery locations across Germany were generated, at a fraction of the cost of a conventional shoot. For WGS Schwerin, a hybrid of AI and live action produced a 30-second corporate film in 27 days, which also ran as a cinema ad at the Filmkunstfest MV.

Planning a campaign? Talk to our AI advertising agency in Berlin.

Note: This article is for general information only and is not legal advice. For advice on your specific situation, please consult a lawyer.