ASA issues a reminder that the CAP Code applies to AI-generated content

This means that the ASA views advertisements in the same way regardless of whether they are generated by an AI system, a human author or a combination of the two.

The ASA goes on to remind advertisers of the need to exercise caution when using generative AI tools to create ad content. The ASA highlights two key risks:

• the risk of an AI system generating misleading claims; and
• the risk of an AI system amplifying biases.

Whilst the ASA notes that it is not aware of having ruled on an ad featuring AI generated content to date, it seems likely that the ASA has focussed on these risks in light of its recent decisions upholding complaints related to socially irresponsible advertising – e.g., those that perpetuate harmful gender stereotypes or portray idealised or filtered body standards.

It is interesting that this statement was published on the same date as the Science, Innovation and Technology Committee’s interim report on governance of AI. That report handily identifies twelve challenges related to AI, and the ASA’s statement reflects two of them: the bias challenge and the misrepresentation challenge, which touches on the impact of AI systems “hallucinating”.  

This is a reminder to users of AI systems of the UK’s current non-centralised approach to AI governance, meaning that industry practitioners need to refer to various sources for guidance on best practices when using AI systems to create content.

Finally, advertisers need to remain vigilant when using AI systems. Human final review remains a critical step when producing content, even in an environment where technology providers are pushing for users to trust the content that their AI systems produce. For example, Microsoft has pledged to defend users of some of its AI products against copyright infringement. Whilst helpful, that's unlikely to be of assistance to advertisers should content created using those AI products fall foul of the risks the ASA identifies.