ASA Upholds Complaint Against Midnite Over AI-generated Character in TikTok Ad

Key Points

  • On 16th September 2026, ASA decided to prohibit a TikTok Ad called “Midnite” for having an AI-generated main character looking younger than 25 years, thus violating CAP Code rule 16.3 and 16.3.14.
  • According to Affiliated Limay Media, no such specification regarding the character’s age had been made in their AI prompt and that the routine approval process with the clients had been overlooked.
  • White Hat Gaming was exempted from the rule in a separate ruling on the same day, after the ASA concluded that the audience on its esports website didn’t comprise the majority under 18.

No one cast him; no one employed him; no one knows where he got his driver’s license from. In the TikTok commercial for Midnite that was banned by the Advertising Standards Authority, the young man who stars in the ad is actually imaginary; created using an algorithm based on a prompt which never once referred to the age of the youth. On 16 September 2026, it was determined by the Advertising Standards Authority that this made no difference.

A World Cup Clip That Set Off the Regulator’s Monitoring System

In its formal decision, the ASA refers to only one complaint about the TikTok ad that was posted on 3 July 2026, and states that the complaint was made in light of the monitoring performed by the regulator through its Active Ad Monitoring programme, which searches for advertisements online that potentially violate the code. In this advertisement, the man used an online gambling service and says: “Another massive World Cup upset. Place a bet of £10 on the big match and receive £30 in free bets. Level up your World Cup with Midnite, get on it.”

The ASA noted that the main character in the video had the appearance of youth, such as young facial features, smooth skin, a teenage haircut, a grey hoodie, and animated facial expressions which all made him look like a man in his late teens or early twenties. The regulator added: “There was nothing in his appearance, styling or behaviour which suggested that he was older than 25 years of age”. This single statement is the operative test in CAP Code rules 16.3 and 16.3.14 which prohibit anyone who is or looks under 25 from participating in any form of gambling and/or gambling advertisement.

The Affiliate Chain and the Missing Approval Step

The response provided by Dribble Media Ltd t/a Midnite to the ASA was that the advertisement had not been approved by them and the brand used was not theirs either, despite the inclusion of the brand “Midnite” in the video. Upon further investigation, it was discovered that the advert was created by their affiliate known as Limay Media, whom they had not engaged to post any paid social media ads on their behalf.

Limay’s account of events revealed a gap that the ASA found difficult to accept. The affiliate said their usual process for gambling advertising required managerial review and approval before publication, followed by client approval, but acknowledged that process had not been followed. They described it as an isolated human error and said the ad had remained online for only a few hours before a manager identified the issue and immediately paused it.

The question of operator liability for unsanctioned affiliate content is not new to UK gambling regulation. What gave this case its particular weight was what Limay’s admission revealed: AI-generated ad content was being produced and published without any age specification built into the creative brief.

Why “AI-Generated” Offered No Legal Defence?

Limay confirmed the character was entirely AI-generated, with no real person used as source material. The prompt used to create the character did not refer to his age or to the age restrictions that govern gambling advertising. The ASA therefore found no evidence that, when generating the character, the AI had been instructed to ensure he appeared to be aged 25 or over.

The stance taken by the regulator is very clear and must be carefully studied. It was never stated that AI-generated ads were inherently a problem; rather, it found that there was a lack of age instructions in the prompt leading to youth-like output and that no external review picked this up before publishing it. Limay had already put in place certain measures, which included age-related prompts, reviews to detect youth age, not using styles and features suggesting people below the age of 25 years, and keeping logs of AI creation and approval.

White Hat Gaming Cleared in the Same Week’s Rulings

The ASA issued another ruling dated 16 September that came to the contrary finding for White Hat Gaming. In this ruling, the ASA considered display advertising for Casimba and The Grand Ivy Casino on HLTV.org, which is an esports site dedicated to Counter-Strike 2. The said display advertisement was conducted in March 2026. The complaint made was that the said advertisements were directed to minors.

White Hat Gaming gave evidence from its referrals proving that most of the visitors who came to the website through HLTV were adults with an age above 25 years. The ASA ruled that the analytical content of the site, match analysis and editorial tone were meant for adults. The complaint was dismissed, as there was no breach of rules 16.1 and 16.3.13 of the CAP Code.

The two cases turned on different legal tests. The Midnite ruling assessed the perceived age of an individual in a paid ad; the HLTV ruling assessed whether a media placement was directed at under-18s. The evidence each operator could produce, or could not produce, was what drove each outcome.

Midnite’s 2026 Formal Ruling Record

TikTok ruling comes as the third ASA ruling against Midnite in 2026. In their decision in August 2026, ASA banned five Midnite advertisements from television, video-on-demand services, and the radio due to 125 complaints. In the commercials, people were struggling with technology breakdowns using the tagline, “We can’t fix the rest, but our betting app is actually pretty good.” The ASA argued that this meant that gambling can be a solution to problems and can be used as a coping mechanism. Another ruling from June was not confirmed.

This follows a period of growth for the company. In January 2026, Midnite managed to secure $35 million in funding as part of its Series C financing deal led by Raine Partners IV and its total equity financing went beyond the $75 million mark. The money was meant to be spent on expansion of its operations, product development, and international growth.

The 4 June 2026 ASA enforcement notice for all gambling operators stated that the monitoring and enforcement will become active from 11 June. According to an update released afterwards, the Active Ad Monitoring system uses artificial intelligence to scan online advertisements, and within four weeks, this system scanned more than 10,000 individual pieces of content, found compliance problems for seven advertisers and all 36 offending advertisements were withdrawn. The Midnite TikTok advertisement aired on 3 July.

Expert Analysis

We think this ruling deserves sharper attention than it has received in trade coverage, and for reasons that go beyond the surface-level “AI is not a defence” headline.

The real issue is not that Limay Media used a generative AI tool. It is that the company treated AI image generation as a black box with no compliance step attached. The safeguards Limay introduced after the ruling, age-specific prompts, perceived-age checks, retained records, are not technically complex. They were simply absent. Any affiliate or operator currently producing gambling creatives with generative AI tools should ask whether their own prompts specify an age at all. The ruling makes clear that if the answer is no, the resulting output will be judged solely on how old the character looks to a viewer.

There is also a liability question the industry has not fully confronted. Midnite terminated Limay’s contract and argued, credibly, that it had not authorised the ad. The ASA directed both parties in its action. That is not a distinction the regulator was prepared to let an operator use as a shield, and operators expanding through wide affiliate networks may find that a harder position to maintain as AI-generated ad production scales. Being unaware of what an affiliate has published is not the same as being free of responsibility for it.