Expert Speak: How to use ASCI’s guidelines on Generative AI for ads while ensuring compliance and ethical conduct

While AI offers advertisers a quick way to connect with users and give a fun twist to their products, many recent examples have flouted advertising guidelines. Advertisers have been finding ways to understand the legal guardrails to navigate AI, which pushed watchdog ASCI to launch a Whitepaper. Social Samosa finds out how it will shape the future of advertising and the next steps.

Last week, ASCI and Khaitan Co released a much-needed whitepaper on Artificial Intelligence (AI). The paper reiterated how advertisements made with AI are subject to the same principles of regulation and consumer protection as any ad. Privacy, copyrights, and responsibility over content creation are some of the critical issues that need to be dealt with in time and ASCI has taken a step forward in addressing these concerns. 

One of the most important guidelines that ASCI reminds us of is consent and copyright violation. 

A few months ago, a song by Drake and the Weeknd called “Heart on My Sleeve” went viral. However, the song’s heart was missing. It was entirely generated by artificial intelligence, courtesy of a TikTokker named “Ghostwriter977” with vocals purporting to be Drake and the Weeknd. It saw over 15 million times before getting pulled from streaming services by Universal Music Group (UMG) after going viral. 

About this infringement of copyright, the whitepaper reminds that it is pertinent to obtain the necessary authorizations and licenses for uploaded materials, including copyrighted and trademarked content.

“In the absence of specific AI advertising guidelines, advertisers can embrace principles from the whitepaper for fair decision-making, accountability, operational transparency, and considering AI's societal impact,” Manisha Kapoor, CEO and Secretary General, ASCI told Social Samosa. 

And this is why the Whitepaper was launched. 

Hybrid ad agency Pulp Strategy has been talking about using Generative AI to craft narratives on Instagram. Its founder and MD, Ambika Sharma, said, “The new guidelines provide much-needed clarity on what is and isn't allowed, which will make it easier for advertisers to use AI in a responsible and compliant manner.”

Brands and agencies haven’t fully adopted and committed to Gen AI as yet, said  Preetham Venkky, Chief Digital Officer, DDB Mudra Group, partly due to there being tremendous confusion around its legal boundedness. 

“Guidelines such as these will help them form internal SOPs and policies for responsible Gen AI usage. With such guardrails, we can see an explosion of Gen AI usage along with appropriate disclosure. The legal precedence and the laws haven’t been formulated yet. This would leave us in Wild Wild West territory. It’s important to see AI as an assistive tool and not a pass-through tool. There will always be a need for proficient and skilled talent who can responsibly use such tools along with the common sense that needs to be applied,” said Venkky. 

Generative AI and its Legal Challenges

The adoption of generative AI, driven by models like OpenAI’s ‘Generative Pre-Trained Transformer’ (GPT), Google’s ‘Bard’, Midjourney Inc’s ‘Midjourney’, Adobe’s ‘Firefly’, is increasing at a quick pace. Generative AI offers a vast range of benefits in terms of efficiency, cost reduction and more. It involves using large language learning models to generate original content based on input data and prompts. 

But as the advertising industry begins to dive deeper into the world of AI, advertisers need to be cognizant of issues around ownership of AI-generated content, data security, inherent AI bias, authenticity of prompts, etc.

Example of probable copyright violation seen in India includes brands like Myntra, Zara and others reimagining Bollywood actors as Barbie. While this was a topical moment and many praised the technology and the brands, there were equal amounts of questions raised on consent and authenticity of the ads. 

“What is unregulated and not covered by legal guardrails, it's not good for business,” reminded Naresh Gupta, Co-Founder & Managing Partner, Bang in the Middle. 

Advertisers, therefore, should evaluate best practices before deploying generative AI for advertising operations.

Giving a tip to brands on regulating AI, Kapoor said, “Advertisers can reduce liability by implementing robust content review processes, setting guidelines, and adding AI disclaimers in marketing materials.”

The paper further ponders upon an interesting dilemma that advertisers could be facing currently. While using Generative AI, it is challenging to identify the exact source or inspiration of a user’s input. However, the paper mentions, if the data provided by an individual is based on or resembles pre-existing copyrighted works, the generated output by the AI could potentially be considered infringement. To illustrate this in the context of advertising and marketing, the paper considers a practical example: 

Suppose we entered the following prompt on an open AI platform: ‘Create a watercolour painting of the Gateway of India in Mumbai, interpreted in the distinctive, swirling, and vivid style of Vincent van Gogh’s painting, Starry Night, with the night sky flowing out into a galaxy.’

When creating an image to promote tourism in Mumbai, an advertiser might use an image that resembles the distinctive style of Vincent van Gogh’s ‘Starry Night.’ However, paper discusses how van Gogh’s artworks are now in the public domain, and the rendered image would not be considered a derivative work of a copyrighted piece. As a result, the advertiser may be entitled to separate copyright protection for the image.

Also Read: The power of marketing lies in the connection you make with your customer: Anukool Kumar, Tinder

The whitepaper also states that the use of generative AI in ads poses risks related to the use of proprietary personal data, the use of publicly available personal information to train AI models and more. 

Earlier last month, Sarah Silverman, acerbic comic, showrunner and author sued OpenAI and Meta, alleging that the tech giants unlawfully used her copyright book to train their artificial intelligence systems.

The guidelines suggest advertisers can consider using gen-AI tools that are trained to reject personal information or simply not provide results when such data is inserted. 

Co-existence: The Way Forward

Closer home, EdTech platform UpGrad faced criticism for featuring an AI-generated person in their ad who had a striking resemblance to Google’s CEO Sundar Pichai. People started questioning the legal and ethical boundaries of consent and credibility. In response to this backlash, the brand then included a disclaimer in the same post that said, “This post is a work of fiction.”

The guidelines shared recently clearly mention the risks that might follow and commenting on that, Mithun Mukherjee, ECD, FCB Kinnect said that brands will have to ensure that they practice due diligence when creating AI-based campaigns. 

Further adding, “If the guidelines are anything to go by, they are just that - guidelines. With the slew of copyright lawsuits that are already being served by Celebrities to AI Generative platforms like Open AI, there is still time before brands and celebrities reach a state of co-existence and not step on each other’s toes. Till then, one can expect lines to remain as blurry as they currently are.”

Also, AI does not have legal status in India, and an AI tool cannot be considered an ‘author’ or ‘owner’ of the works it creates.

Akshay Gurnani, CEO and Co-Founder, Schbang said, "Having read through the report it's clear that AI is not recognized as a legal entity in India and hence cannot be regarded as an 'author' of the content being produced. This also means that agencies will have limited recourse in case of infringement by a third party. It's also important to ensure that none of the prompts are created with the intention of generating infringed or copyrighted content. This leaves the entire onus of using AI responsibly on the agency or brand and it's a wait and watch until more stringent regulations are put in place.” 

View this post on Instagram

A post shared by Social Samosa (@officialsocialsamosa)

As advertisers better understand legal guardrails, Mukherjee said that brands will have to ensure that they practice due diligence when creating AI-based campaigns. 

“If the guidelines are anything to go by, they are just that - guidelines. With the slew of copyright lawsuits that are already being served by Celebrities to AI Generative platforms like Open AI, there is still time before brands and celebrities reach a state of co-existence and not step on each other’s toes. Till then, one can expect lines to remain as blurry as they currently are,” said Mukherjee.

He also advised advertisers and marketers can go by one simple thumb rule and that included staying away from non-licensed celeb-related AI generation. 

Gan.ai, which is the agency behind recently released AI campaigns such as Pepsi’s Friendship Day campaign, Cadbury Celebrations’ #MyBirthdaySong campaign, Jeevansathi’s personalized wedding invites and others make sure to get their clients to obtain the informed consent of the actor in their personalized videos mentioned Anupreet Singh, Chief Revenue Officer, Gan.ai.

He also mentioned, “Consumers today demand personalization that can only be achieved through AI. Thus, mitigating bias is not only a guideline for safety and ethics but also a necessity for improving AI adoption and application at scale.” 

He further added that the country has just scratched the surface of what is possible with AI, and the winds of change have not slowed one bit.

As more tech companies jump on the AI bandwagon, the industry could move beyond the paywall, bringing more clarity. 

Commenting on the current state of AI in the country, Naresh Gupta stated, “Right now it's all free for all, and that is not good for either the creative community or the marketing community. Eventually, the creative product has to be owned by the entity that pays for it, but that ownership today in its current form is opaque.”

And as far as the consumers are concerned, a recent report found that found 77% of consumers believe positive customer experiences still need an element of human touch, and 58% of consumers say they want companies to be clear about when AI is being used. 

To win consumer trust, the future of AI and advertising will see all stakeholders co-existing. 

“AI is a game-changing tech with profound implications for advertising. As the field evolves, privacy, data protection, transparency, and responsibility challenges need focus to ensure consumer welfare in this emerging landscape,” said Kapoor.

It is only through collaboration that the advertising industry will build trust and maximize the benefits of AI extensively while following responsible AI practices.