ASA warns advertisers over AI content and deepfakes

TL;DR:

• The Advertising Standards Authority has clarified that its existing rules apply fully to AI-generated advertising and deepfakes.
• Brands remain responsible for misleading, harmful or unsubstantiated claims, regardless of whether AI produced the content.
• The guidance lands as deepfake scam ads and AI-made imagery proliferate across UK platforms.

The UK’s advertising watchdog has put marketers on notice that AI offers no shelter from the rules. The Advertising Standards Authority has set out its position on AI-generated content and deepfakes, making clear that the existing advertising codes already govern such material and that brands carry the same liability as for any other ad.

No new rulebook, no new excuses

The ASA’s central message is that it does not need bespoke AI regulation to act. Whether an image, voice or claim is produced by a person or a model, the principles are unchanged: advertising must not mislead, cause harm or make claims that cannot be substantiated. That puts the onus on advertisers and agencies to check AI outputs before publication — a non-trivial task when generative tools readily invent statistics, fabricate endorsements or render photorealistic scenes that never happened.

Deepfakes draw particular concern. The ASA flagged the growing problem of fraudulent ads that clone the likeness of public figures to lend false credibility, a tactic increasingly used in investment and crypto scams. For legitimate brands, even consensual use of synthetic faces or voices raises disclosure and consumer-trust questions the codes are designed to police.

The intervention fits a wider pattern of UK regulators racing to apply existing powers to AI rather than wait for new statute. It echoes the FCA’s warning that AI is making financial crime “cheap, fast and invisible” and the ICO’s plan for safe AI innovation — each signalling that compliance obligations already bite.

Looking forward

For UK businesses, the practical takeaway is governance, not novelty. Marketing teams experimenting with generative tools need review processes that catch fabricated claims and unlabelled synthetic media before campaigns go live. The ASA has signalled it will treat AI-driven breaches as it would any other, so the cost of a careless prompt is the same reputational and regulatory exposure that has always followed a misleading ad — only now the failure can be generated in seconds.