Avoid the risks and reap the rewards of using AI in advertising

The advertising industry has seen massive transformation in the last five years. The shift towards online advertising revolutionised the industry and now artificial intelligence (AI) is set to have a bigger impact on advertising that any other industry. Our IP and Media team explain how to make the most of the opportunities AI offers in advertising, while keeping an eye out for possible intellectual property pitfalls. 

The opportunities

Lower costs: By creating content quickly and cheaply, generative AI makes it possible to have a more direct relationship with consumers and to create content that previously would have been too time-consuming. 

Accessible technology: Agencies worried about tech companies offering ad generation direct to brands can embrace generative AI too. 

Better media planning: AI can sift through vast amounts of data to inform where and when to place ads, for example tracking the weather to predict when people are more likely to buy seasonal products. 

Crunching large data sets: AI can spot trends to help develop new products or shape ads.

Hyper-targeted personalised ads: AI could potentially be used to deliver ads that reach the right person, at the right time, with the right message and in the right context. 

The risks

Increased monitoring: The ASA is using AI to check 10 million ads in 2024, targeting greenwashing and ads that promote gambling and vaping to children. 

Lack of UK legislation: It’s clear that copyrighted material is used to train AI; whether that constitutes an infringement of the copyright of those works hasn’t yet been dealt with by the UK courts. Getty Images V Stability AI is one of the first opportunities to see how they deal with this. 

Content ownership: Who owns the IP of AI-generated content affects how you can exploit the content. Platforms generally make it explicit that the user retains ownership of the inputs and that you own the output but check. 

Copyright infringement: Be aware that AI-generated work might reproduce the whole of, or a substantial part of, a third party work. If you’re developing an in-house AI model, obtain licenses for the datasets you’re using to train the model. 

International campaigns: There are differences in how countries are regulating AI, so seek local advice for international campaigns. 

IP infringement: There are some high profile examples of trade marks working their way into AI-generated content because the trade marks are in the training data. Look out. 

Check rights: If using a celebrity image, it's vital that you have the necessary rights or there’s a risk of false endorsement. 

Data protection: Personal data used in AI training data can lead to issues of breaches of the relevant data protection privacy laws. The Information Commissioner’s Office (ICO) is consulting on how aspects of data protection law should apply to the development and use of generative AI models. Watch this space.

Beware bias: Biases can creep into the training datasets. It may be difficult to review all of the outputs AI is creating, so try to guard against AI bias.

Advertising codes: Make sure that your ad complies. Using AI-generated material to substantiate advertising claims can build in risk; consider the reliability of the datasets and the prompts you’ve provided.

AI and advertising an evolving area. 

These five tips will help protect your reputation:

1. Make sure you understand whether AI has been used and how.
2. Review the terms and conditions of the AI service provider: Who owns the rights? Where do the liabilities lie? Is there an indemnity for any third party IP infringement claims against you arising from use of the AI-generated content? What about data protection?
3. Make sure that any advertising claims stack up.
4. Have processes in place to deal with complaints or data protection queries efficiently.
5. Train relevant staff on the issues and risks. Have policies in place that staff must disclose when content has been generated by AI or AI-assisted so you can manage that risk.