Advertising and AI: time for the label

On TikTok, in Nigeria, an ad in Pidgin announces a grant. The OPay logo is there, the voice is warm, the link promises money within minutes. None of it is true. The voice was made by an AI, the brand asked for nothing, and the link leads to a page that asks for a phone number and bank details. Between April and May 2026, journalists at Premium Times archived 130 ads of this kind in five weeks: OPay, UBA, MTN, Airtel. They spoke English, Pidgin, Yoruba, Hausa. None carried the words “AI-generated”.

Fraudsters learned the language of our audiences before many brands did. And they understood before the brands did that AI makes it possible to speak to everyone in their own language, quickly and for next to nothing.

On 30 September, Nigeria raised the question

On 30 September 2026, Nigeria’s consumer protection authority, the FCCPC, put draft regulations on commercial promotions out for consultation. A whole chapter deals with artificial intelligence and automated marketing. What it proposes comes down to four ideas.

Any business that uses AI or automated tools for its advertising or customer relations would have to declare it to the regulator. Content produced by an AI would have to be clearly flagged, so that consumers know they are talking to a machine and not to a person: the text cites chatbots, virtual influencers and automated messaging. The business would answer for everything its system says, even when the system acted on its own: no shifting the blame onto the machine. Finally, consumers could refuse automated messages.

The proposed penalties are heavy: up to 100 million naira or 1% of the previous year’s turnover, whichever is higher, and, for directors, a ban of up to five years from managing a company. The consultation closed on 20 October. So this is a draft, not yet a law: it may change, and it will only apply once published in the Official Gazette.

The text does not come out of nowhere. In April 2026, the same authority had already published a note setting out how the 2018 law on deceptive practices applies to AI-generated content: fake testimonials, fake reviews, deepfakes, chatbots, retouched product images. And on 15 September, ARCON, Nigeria’s advertising regulator, condemned a fake Facebook ad in which AI-generated images showed President Tinubu and Aliko Dangote promoting an imaginary investment fund.

Fake ads move faster than the platforms

On 8 October 2026, Premium Times published a second investigation, this time on Facebook. In it, sponsored deepfakes impersonate Ngozi Okonjo-Iweala, Tony Elumelu, Jim Ovia or Aliko Dangote to sell fake investments. One of these ads, sponsored in mid-April, was only taken down in July, after the newsroom contacted Meta.

French-speaking Africa is not spared. In March 2026, a doctored video cloned the image and voice of President Alassane Ouattara to recommend an investment platform.

The platforms are moving. Since February 2025, Meta has labelled ads created with its own AI tools, and since 1 June 2026, those it detects as made with third-party tools. TikTok requires its creators to label realistic AI-generated content, and says it has labelled more than 1.3 billion pieces of it. But the fraudulent ads in both investigations ran without a label. Detection relies on markers that fraudsters have no reason to add.

The label protects honest brands. It does not stop scammers.

French-speaking Africa is not on the sidelines

We often picture these debates as far away from us. They are already here.

In Gabon, an ordinance of 26 February 2026 bans deepfakes that depict an identifiable person without their consent, and requires platforms to mark AI-generated content visibly. In Côte d’Ivoire, the Conseil supérieur de la publicité announced in June 2026 stronger regulation of digital advertising and rules for influencer marketing; the same month, in Abidjan, the Minister of Communication warned against deepfakes and called for Africa to write its own rules. On 6 July, six West African countries, including Benin, Côte d’Ivoire and Senegal, presented joint guidelines for ethical and responsible AI, which are not binding. In Senegal, the new Conseil national de régulation des médias also covers content creators.

In Cameroon, we found no rules on ads made with AI. Yet the Ministry of Posts and Telecommunications devoted its 2026 Digital Innovation Week to the abuses of AI, deepfakes included. The subject is on the table.

And for brands that speak to the diaspora, Europe has already decided: since 2 August 2026, the EU AI Act has required those who distribute deepfakes, advertisers included, to disclose it. In France, the advertising self-regulatory body (ARPP) published a simple grid in August for deciding when to label: the level of AI involvement, the degree of realism, and the risk of misleading people about authenticity.

The public likes AI. It hates being deceived

It would be a mistake to use AI to frighten people. African audiences are among the most open in the world. In Nigeria, according to the global survey by KPMG and the University of Melbourne published in 2025, 79% of people are willing to trust AI. But 61% cannot tell whether online content is reliable, since it may have been generated, and 88% want laws against AI-generated misinformation.

A US survey published on 20 October 2026 gives an idea of what hidden AI costs: 74% of respondents believe brands should disclose AI in their ads, and 44% would trust a brand less if it had not. AI-generated people are what the public rejects most in those ads. These are American figures, not to be transposed as they stand. No measurement exists yet for French-speaking Africa, and that in itself is telling.

Research, for its part, has not settled the question. Some studies find that an “AI-generated” label lowers trust in the ad, others that it strengthens it. What we do know, on the other hand, is what happens when the public finds out on its own what was hidden from it.

Two brands, two answers

In December 2025, McDonald’s Netherlands released a Christmas ad generated entirely by AI. Three days later, it was withdrawn, after a wave of criticism of images judged disturbing.

In June 2026, in South Africa, Windhoek beer took the opposite approach. Its campaign features a TikTok creator born with six fingers on each hand, a nod to the botched hands of AI-generated images. The brand committed never to show AI-generated humans or beers in its advertising, and to have all its partners respect that commitment. It turned authenticity into a selling point.

Between the two lies the path most brands take: use AI where it helps, and say so.

What we do with it at mystiQ

We produce with AI: images, voices, videos. We teach it too. On our site, every AI-generated image carries the words “AI-generated image for mystiQ.”, and no generated face is presented there as a client or a team member. That is the rule we are extending to all our content, social media included.

AI does not frighten us. What worries us is the brand that hides it, and the brand that lets fraudsters speak to its customers better than it does. The TikTok scammers spoke Pidgin, Yoruba and Hausa. They had done the work we call cultural intelligence: learning the language of their targets before speaking to them. A brand that wants to protect its customers must do at least as much, and warn them in the languages in which they are being deceived.

So, one last question: if one of your customers found out tomorrow that one of your ads had been made by a machine, would they hear it from you, or from someone else?