The Future of Advertising 2024: AI Legal Update

The conversation surrounding legal issues with using AI content have developed rapidly since last year. In line with this, when looking ahead at The Future of Advertising during this year’s event, the APA had to get into these changes. 

Steve Davies, CEO at the APA was accompanied by Jamie Smith, partner and co-head of the Interactive Group at Sheridans on stage and together they walked through numerous questions regarding the elusive legality of generative AI – copyright infringement, AI usage infringing rights of third parties, conflicts between obligations under the standard production contract and usage of AI and much more.

When we create content using AI, the data has been taken from somewhere else – a point that has been already well-established and Jamie believes the industry should be building upon it now. “But what does that mean for ownership,” he asked.

“When it comes down to confidentiality, there is so much out there that you can upload – scripts, ideas and more – it’s important to recognise these are live environments in a lot of cases. As soon as you do that, they’re gone,” he reminded the audience.

“If you think about NDAs and similar contracts, you have to ask yourself if you even should be uploading those. The content is being reinvested. These ideas, characters, pitch decks are in some way working their way back through the system.”

In this sense, could AI be an impediment to creativity if it’s just recycling already existing material? Jamie believes that ad folks should really think about the ways in which it can help them, rather than simply asking it to “create something” and taking it away.

As Steve reminded, when it comes to re-ingesting content the issue isn’t only about IP, but also about a break of confidentiality obligations towards clients. And to that point, Jamie was categorical that if something does go wrong, one should never assume that their insurance will respond. 

“They should do, but they will most definitely look for the sensible steps you have taken to control how these platforms have been used.”

A very important point Jamie wanted to touch on was also content clearing – you should never assume the content is fine because you used any particular system, you need to have a form of content clearance going on in any case, he stressed. “If anything, agencies are looking more and more into undertaking deeper processes to take more of that clearance on. Platforms are not creating, they’re just responding to your prompt, not undertaking any clearance themselves.”

Especially if you’re concerned with a live environment, for example a promotion happening during a live event, where you’re allowing the system to respond in real time is when you should ask yourself ‘What safety nets have we got in place?’. “And again, the answer might be in the contract. Who’s ultimately responsible if this goes wrong? Should it be you, deploying the technology? Or should the risk be shared and explained to the end brand,” said Jamie.

A significant issue in the contracts at the moment is that, as Steve reminded, you are essentially “giving a guarantee that if you’re using AI you may not be able to fulfil the contract.

“What this means is that if you are doing an AI project, you have to explore the practical risks with the agency and talk about a sensible sharing of those risks, and not requiring you to sign contracts that you’ll immediately be in breach of,” said Steve. “As long as you did that and as long as the risks are explored, it should be entirely possible to get there, because a blanket ‘no’ from the agency to assume any part of the risk would just make it impossible for you to do the project without being in full breach of obligation.”

A potential way forward, according to Jamie, would be to try and restrict further use of the content you create by AI and clients, which ultimately goes back to the clearances point. “If you’re creating something that can be ingested and reused, they’re essentially removing you from the value chain in a way,” he said. “So again, it’s this sort of control over the content which makes clients come back to you.”

Looking back at the SAG-AFTRA strikes of 2023, Steve said that stating clearly that the principle the industry should follow is: “You can use our stuff as intended for you and can use it only once, but you cannot use it forever through various AI systems without paying us or giving us any credit.” But it is also a principle which the current contractual situation does not reflect, “as [the contracts] were drafted prior to the widespread use of AI”.

So what is the industry’s position today? Jamie pointed to the evidence that the tools are being used increasingly since last year in various parts of the production pipeline and some to fully create content from scratch. The noise around high-profile cases regarding AI has also dissipated a bit, which has allowed the industry to focus on the way the tools are being used. 

Steve also grounded the conversation by reminding that while hypothetical risks do exist, infringing copyright by using AI poses, in reality, an almost negligible risk. “To establish an infringement of copyright, you need to establish that it is a copy of the work – at a very basic level it has to look like the ‘thing’ you’ve copied. And if an AI tool is using millions of images to create something that doesn’t look like anything, then no, there’ll be no claimant.”

Remaining realistic is important too, said Jamie when explaining that not everybody is in a position to be seeking advice when things approach grey areas. However, when clients require usage of AI or application of AI processes to creative campaigns, is when a much more layered risk emerges. 

“In that sense,” said Jamie, “It is not unreasonable to explain that if you were to take this risk on, you do need to include some costs which incur external advice, or that may just cover the team internally, in terms of their additional time in running through the licences, clearances and consents.”

He continued: “If they say ‘no’ you can ultimately push that risk back onto them and say that you cannot be responsible for the final clearance. We must be aware of this and try to push that approach, because increasingly you’re going to get more instances where the responsibility and the risk will be on you for the same job value.”

As these issues aren’t easily resolved and don’t have a single answer, they’re constantly changing along with each contract draft and use case. The way the industry looked at these conundrums a year ago is completely different to the way they’re perceived today, agreed Steve and Jamie – it is something that should be looked at on a continual basis, but not something over which extra costs and risks should be taken.