What's the issue?

Small print or footnotes are commonly used in advertising to qualify more prominent claims or images. Here, we recap existing rules around the use of small print and consider how the use of AI might impact what is disclosed, focusing on the requirements under the ASA's CAP Code. 

Small print – what are the basics?

The basic rule under the CAP Code is that marketing communications must not materially mislead or be likely to do so. When applying qualifying information, the information must:

• Be included where it is material to the consumer’s understanding of the primary claim, or marketing communication as a whole. 

• Not contradict the primary claim being qualified to the extent that consumers are likely to be misled.

• Be presented clearly with an appropriate level of prominence.

As such, qualifications are generally used to ensure that the ad, as a whole, does not mislead the consumer, clarifying the context of the ad rather than contradicting the main or obvious message. 

Calling it 'small print' is perhaps misleading in itself as ad qualification or clarifying language must be presented clearly and with sufficient prominence. If the information is material to a consumer's understanding of the product or service on offer, it should be presented so that it can be easily seen and read by the average consumer. The qualification may need different levels of prominence or detail depending on the context, whether it be a product or service being advertised, the target audience, and media in which the advert appears.

The ASA has issued guidance on the use of qualifications in ads which provides helpful information on how and when qualifications should be used. There is similar guidance on the use of superimposed text in broadcast advertising.

How has the CAP Code been applied to small print?

The ASA upheld a complaint against 'Screen With Envy Ltd' relating to qualifications, where the ad stated "20% off all orders over £200", reinforced by using "sitewide" and "all orders" as qualifications alongside the main claim. The ASA considered this to mean that, when the consumer used the discount code provided, they would receive a 20% discount on any order over the £200 threshold. The ad failed to make clear that the discount was not valid in conjunction with other discounts and was therefore misleading. Even if the ad had made this clear, the small print would have contradicted the main claim rather than serving to clarify it.

There are a number of examples of ASA rulings against mobile or app-based games where the ad fails to accurately represent the actual gameplay, or where the gameplay featured lies behind a paywall or is only available in the 'premium' version of the game (requiring the user to subscribe). Typically, the use of a qualification such as "Not all images represent actual gameplay" will not be sufficient to overcome any objection that the advert is misleading as users will play a significant amount of gameplay which is different in style or quality to access the gameplay featured in the ad (see, for example, the ASA ruling on PLR Worldwide Sales Ltd t/a Playrix).

How does AI affect the position?

Whilst the current landscape for the use of qualifications in ads is quite clear, the encroachment of artificial intelligence (AI) into various aspects of advertising may mean that closer thought has to be given to how and what is disclosed or qualified.

Do AI-written plots or storylines need disclosing? How about AI-generated images in advertising? Or AI-generated video? What if the content is dynamic such that it evolves depending on consumer response or engagement?

What has the ASA said on the issue?

The ASA sought to clarify the position last year and reiterated that 'the Code is media-neutral'. If an ad falls under the ASA's remit, the existing rules will apply, irrespective of whether the ad was created (in whole or part) using AI.

That doesn't mean that the use of AI in the creation or development of the ad can be ignored. The basic rule and consideration of whether the ad will materially mislead the consumer or be likely to do so is paramount. Thus, using an example provided by the ASA, if AI images are used to make efficacy claims (similar to photoshopping or filters in social media), the advertiser should give consideration as to whether the overall effect is one of misleading the consumer as to the efficacy of the product. So, in some instances a qualification or 'small print' may be necessary.

In short, it is not necessary (yet) to disclose that part or all of an advert has been generated using AI, providing the advert is not misleading and does not otherwise breach the CAP Code. However, care should be taken to ensure that any images and text created do not mislead.

What other issues does AI generate?

As an adjunct to this, the ASA separately highlighted an issue whereby using AI in advertising may potentially lead to socially irresponsible ads. The way AI models are trained (using large underlying data sets) may amplify biases present in the underlying data set and could lead to, for example, AI-generated ads showing women as the primary caregiver, idealised body standards, or individuals of a particular race performing certain jobs. This must be guarded against.

More complex use of AI to create deepfakes, rather than still images, is still relatively new in advertising and (to date and to its knowledge) the ASA hasn’t issued any rulings against ads featuring deepfakes. However, the ASA has issued some guidance to advertisers who might be considering deploying this technology in their ads in order to stay on the right side of the Code. In particular, the ASA states:

• Err on the side of caution – aside from any regulatory hot water you might end up in, you could well get in trouble with the public figure you’ve decided to use in your ad.

• Personality rights and Intellectual Property are not issues for the ASA, but it’s always worth considering any potential legal action you may be opening yourself up to with your advertising.

• While the ASA only regulates legitimate advertising, the ASA Scam Ad Alert System could come into play if they receive reports of ads using deepfake technology for nefarious purposes.

It seems clear that particular care should be taken before using deepfakes of real people. 

The key message from the ASA is that we can't entirely hand over the reins to AI just yet. Irrespective of whether AI is used or not, the primary responsibility for ensuring the ad complies with the Code remains with the advertiser. Human scrutiny and oversight of AI generated adverts is key. 

What are the future trends?

The ASA has already alluded to the fact that the regulatory framework should develop, enhancing transparency and accountability, particularly for online ads. It's currently unclear whether this means potential amendments to the CAP Code. For now, the ASA is of the view that no changes to the CAP Code to specifically address the use of AI in advertising are needed. Whether that will continue to be its position remains to be seen.

Meanwhile, the ASA is itself using AI to help in online ad monitoring, with automated monitoring algorithms scrutinising advertisements, swiftly detecting non-compliance. That is likely to mean that non-compliant adverts are quickly detected, and action taken.