Synthetic Media: What It Is and Why Advertisers Should Care

A dead singer recorded a new duet with her daughter. Shah Rukh Khan personally recommended thousands of tiny corner shops by name. Jennifer Lopez malfunctioned on camera so you could take over and finish her cruise ad yourself.

None of those things happened the way they look. All three were real, paid campaigns. That gap between "looks filmed" and "was generated" is the whole story of synthetic media, and it is the reason advertisers can no longer treat this as a novelty.

Here is the part most glossary articles skip: knowing the definition is useless. Knowing which uses will win you awards and which will get you sued or dragged online is the actual skill. This article gives you a risk map, a go/no-go rule, and a disclosure checklist you can use before you approve anything.

What synthetic media actually means

Synthetic media is any content created or altered by AI so it reads as authentic. Text, images, audio, video, or a blend. The term is deliberately neutral. It covers the harmless (an AI-generated product shot) and the dangerous (a cloned voice saying things the person never said).

For advertisers, the useful definition is narrower. Synthetic media is media where the "camera" was a model, not a lens. Sometimes that means a fully generated scene with no real people. Sometimes it means a real person's face or voice recreated by a neural network. Those two are wildly different in risk, and lumping them together is how brands get burned.

A quick vocabulary check, because these words get mixed up:

• Generative AI is the technology. Synthetic media is the output.
• A deepfake is a subset of synthetic media: a convincing likeness of a specific real person doing or saying something they did not.
• AI-assisted means humans directed the work and used AI for parts of production. Fully AI-generated means the visual came out of a model from a prompt.

That last distinction matters for your creative team. Most strong ad work today is AI-assisted, not fully autonomous. The human is still the director.

Why advertisers should care right now

The volume is already here. According to IAB data cited in Hedra's breakdown of AI-generated advertising, 85% of advertisers using AI creative deploy it most heavily on social, followed by display at 73% and TV at 56%. This stopped being an experiment. It is a production line.

Three things changed at once:

Cost collapsed. A concept that needed location scouting, a shoot, and weeks of VFX can now be produced from a prompt and a reference image in days. That does not make it good automatically, but it removes the money excuse for testing more creative.

Personalization got real. The same spot can now show a Brooklyn skyline in New York and a different skyline in Chicago, generated on delivery. One brand ran a Super Bowl-scale ad that adapted its backdrop to the viewer's city. Static, one-size creative now looks lazy by comparison.

Likeness became a product. You can license a face, clone a voice, and even bring back a public figure who has died. This is where the upside and the legal exposure both live.

If your competitors are shipping ten creative variations a week and you are shipping one, synthetic media is not a curiosity. It is the reason for the gap.

The four tiers of synthetic media risk

Not all synthetic media carries the same risk. Treat it like a ladder. The higher you climb, the bigger the reward and the harder the fall. Here is the map I use to sort any concept before it goes into production.

Most of the panic around synthetic media is really about Tier 4 leaking into public perception. Most of the actual value for advertisers lives in Tiers 1 through 3, where the risk is manageable if you handle rights and disclosure properly.

What the winning campaigns got right

The campaigns people remember were not reckless. They were deliberate about consent and framing.

Volkswagen's Brazilian anniversary ad brought back singer Elis Regina, who died decades ago, to duet with her real daughter. It worked because it was done with the family and framed as tribute, not deception. Tier 4, handled with the rights holders in the room.

Cadbury's Diwali campaign in India cloned Shah Rukh Khan's face and voice so thousands of small shops could feature "his" personal endorsement of their store by name. It worked because Khan consented and the tool made local businesses the hero. Tier 3, with a clear consent chain.

Virgin Voyages built a spot where a synthetic Jennifer Lopez openly "malfunctions" to reveal the actors behind the avatar, then invited viewers to make their own version. It worked because it was honest about being fake. It turned the disclosure into the joke.

Mango generated an entire fashion campaign for its teen line with no photoshoot at all. Tier 1, low drama, pure efficiency, running across dozens of markets.

The pattern is not "use AI." The pattern is: match the tier to the consent you actually have, and never make the viewer feel tricked.

The mistakes that get brands in trouble

Same technology, opposite outcome. Here is where it goes wrong.

Using a likeness without a deal. If a face or voice belongs to a real person, you need a signed agreement before it touches a timeline. "It's just a lookalike model" is not a defense your legal team will enjoy explaining.

Skipping disclosure when it matters. If a reasonable viewer could believe a real person endorsed you, or that footage is real when it is generated, say it is synthetic. Regulators in multiple markets are moving toward mandatory disclosure for synthetic content, and the court of public opinion is faster than any regulator.

Treating "generated" as "finished." A prompt gives you a draft, not a campaign asset. The ads that flop are usually the ones nobody art-directed, quality-checked, or tested. Synthetic does not mean skip the craft.

Ignoring bias and artifacts. Generated scenes can produce off-brand imagery, distorted hands, or unintended stereotypes. Someone has to review every frame with a critical eye.

The disclosure decision rule

When do you need to tell the viewer content is synthetic? Use this. If you answer yes to any of these, disclose it clearly:

• Does it show a real, identifiable person saying or doing something they did not actually do?
• Could a reasonable viewer mistake generated footage for real events or a real endorsement?
• Does the platform, market, or ad network require it?
• Would the campaign feel like a betrayal if the audience found out afterward?

That last question is the honest one. If the answer is yes, you already know. Disclosure is not a legal chore, it is trust insurance. The Virgin Voyages ad proves you can make the disclosure part of the entertainment.

How to actually use this without getting burned

You do not need to become a rights lawyer or a prompt expert. You need a process and people who know where the landmines are.

Start in Tier 1 and 2. Generate scenes and use synthetic presenters that resemble no real person. Low risk, fast learning, real cost savings. Prove the workflow before you touch anyone's likeness.

Test before you scale. The whole point of cheaper production is more shots on goal. Launch a handful of variations with equal budget, read the results after a week, then double down on winners. If you want a structure for that, our guide on ad creative testing lays out the method.

Get the rights in writing for anything Tier 3 or higher. Scope, channels, duration, and territories. If you are unsure what you are allowed to do with generated footage of a real person or a licensed avatar, read our field guide to AI video rights and licensing before you brief anyone.

Keep a human director in the loop. The best synthetic media still has a person making creative calls, checking frames, and owning quality. If you are evaluating avatar-led work, our AI avatar quality checklist shows what "campaign-ready" looks like versus a rough draft.

This is also the honest case for working with vetted creators instead of raw tools. A campaign-ready AI video creator handles the rights paperwork, the disclosure judgment, and the frame-by-frame craft, so you get the upside of synthetic media without inheriting the risk. That is the gap a marketplace like Viralix exists to close: brands come with a brief, creators come with the production discipline.

The short version

Synthetic media is media generated by AI instead of captured by a camera. For advertisers it is already a production line, not a novelty, because it collapsed cost, made real personalization possible, and turned likeness into something you can license.

Do three things and you will stay on the right side of it:

• Match the tier to the consent you have. Tier 1 and 2 are safe to move fast on. Tier 3 and 4 need signed rights and legal sign-off.
• Disclose whenever a viewer could feel tricked. Treat it as trust insurance, not a chore.
• Keep a human in the director's chair. Generated is a draft, not a finished ad.

Get those right and synthetic media is the cheapest creative advantage on the table. Get them wrong and it is the fastest way to a lawsuit or a public apology.