Boost Innovation and Consumer Trust with AI: The Responsible Path Forward
Mary K. Engle, Advisory Council Member, Executive Vice President, Policy, BBB National Programs
For the past five years, I have participated in the International Council for Advertising Self-Regulation (ICAS), a global platform for advertising self-regulation organizations (SROs) from around the world. Both ICAS and BBB National Programs’ Center for Industry Self-Regulation (CISR) share a core belief: that self-regulation can effectively address industry-wide challenges while promoting responsible business practices through leadership and collaboration.
The work being done at the global level through ICAS provides a valuable lens through which we, at CISR, conduct our work at the national level. Last year, ICAS launched a new initiative, its Global Think Tank, and this year that group, in which I have had the honor to participate, released two thought pieces exploring the ethical, regulatory, and practical implications of using AI in advertising: AI in Advertising, authored by Professor Steven Sidley, and Beyond Simple Labelling, authored by Konrad Shek.
Sidley’s paper reviews several challenges that the use of AI in advertising raises, including transparency, hypertargeting, and dark patterns. Shek focuses on the question of transparency and whether labeling AI-generated content is needed to enhance consumer trust.
Sidley describes two unique challenges AI presents for advertising: First, it is autonomous, evolving on its own, introducing unpredictability; and second, it is opaque, meaning we often can’t see how an AI system arrived at its output. These properties make AI both powerful and harder to regulate and trust.
As a result, Sidley argues, advertising’s ethical north star must remain consumer trust — the foundation of every successful campaign.
There is no doubt that transparency and trust generally are important to ethical advertising, and in his paper Sidley explores a few important challenges that the use of AI in advertising may present to consumer trust.
Sidley suggests a compromise: a simple disclosure such as “Some AI-generated elements may have been used in the creation of this advertisement.”
Sidley urges the advertising industry to counter these risks through independent algorithmic audits, clear ethical design standards, consumer education, technical countermeasures, regulatory approaches, and whistleblower protections. Without vigilance, he cautions, the abuse of AI could erode digital trust and provoke severe regulatory backlash.
Sidley closes with a philosophical question: Should we even try to regulate AI in advertising? His answer is yes — but carefully. Broad “AI laws” are too blunt. Instead, regulation should focus on specific use cases where deception or harm is likely.
His roadmap for self-regulators and policymakers includes:
The goal is not to hinder innovation but to ensure it serves consumers ethically and transparently.
CISR shares this philosophy. In our recent publication examining the risks of generative AI, among our recommendations for industry leaders is to ensure that AI enhancement does not mislead as to product characteristics, function, or performance. Or more pointedly, it is the responsibility of advertisers to ensure that the ad does not mislead or blur the distinction between what is real and what is imaginary.
Shek examines the ethical, legal, and policy implications of AI-generated content in advertising, arguing that mandatory labeling is too blunt an instrument to ensure transparency or consumer protection. Instead, Shek calls for a risk-based, context-driven framework that balances innovation, accountability, and consumer trust.
Shek explores how generative AI may differ from traditional tools such as CGI and Photoshop. He concludes that due to gen AI’s accessibility, speed, and ability to produce hyper-realistic synthetic content without human skill, it may pose greater risks than other technologies. Key issues include:
Despite these risks, Shek warns that technology-based, top-down regulation may be too rigid. Instead, he advocates for use-case-specific oversight. Moreover, existing consumer protection and advertising laws already prohibit misleading or deceptive content globally. The question then becomes whether, and if so, what, additional regulation may be needed for AI-created ads.
Public opinion, too, is nuanced. Shek reviews the available evidence regarding public perception of AI-generated ads. Surveys show strong public demand for transparency but persistent skepticism about AI-generated ads:
Despite the superficial appeal of requiring labeling of AI-generated ads, Shek notes that it is probably not the right approach and risks backfiring for several reasons.
First, Shek points to research showing an “implied truth effect,” where when some content is labeled and some is not, consumers tend to distrust the labeled content even though there is no relationship between the accuracy of the content and whether it is labeled. Second, the existence of AI aversion means some consumers distrust AI-labeled content even though, again, both labeled and unlabeled content may be accurate, or not. Third, given that nearly all ads may soon use some AI, universal labeling could become meaningless or ignored.
In contrast to Sidley’s compromise solution of a disclosure that ad content may have been AI-generated, Shek concludes that a more nuanced approach to the issue is required. He proposes a risk-based framework:
Shek concludes that AI labeling should not be universal but contextual and proportionate, applied only when deception risk exists. The guiding principle remains that advertising must be legal, honest, and truthful, regardless of the technology used. Effective ways forward, therefore, go beyond simple labeling toward smarter, adaptive, and context-driven approaches that consider the actual risk of consumer deception.
Our AI Working Group, announced just this week, is an expansion of a convening that has been active for the last 15 months under BBB National Programs’ Children’s Advertising Review Unit (CARU), resulting in the development of the Generative AI & Kids Risk Matrix.
The CISR AI Working Group will consider the array of harms that AI chatbots pose when children and teens use them as companions and will consider what safeguards industry should consider to protect against these harms while allowing for the benefits.
AI has transformed advertising with breathtaking speed, reshaping creative design, targeting, and consumer engagement, often outpacing the development of guardrails and governance. The collaborative work of the ICAS Think Tank and CISR’s AI Working Group is essential to ensure innovation moves forward responsibly.
We invite industry leaders, policymakers, and researchers to join us in shaping a future where AI innovation and consumer trust grow together.
For the past five years, I have participated in the International Council for Advertising Self-Regulation (ICAS), a global platform for advertising self-regulation organizations (SROs) from around the world. Both ICAS and BBB National Programs’ Center for Industry Self-Regulation (CISR) share a core belief: that self-regulation can effectively address industry-wide challenges while promoting responsible business practices through leadership and collaboration.
The work being done at the global level through ICAS provides a valuable lens through which we, at CISR, conduct our work at the national level. Last year, ICAS launched a new initiative, its Global Think Tank, and this year that group, in which I have had the honor to participate, released two thought pieces exploring the ethical, regulatory, and practical implications of using AI in advertising: AI in Advertising, authored by Professor Steven Sidley, and Beyond Simple Labelling, authored by Konrad Shek.
Sidley’s paper reviews several challenges that the use of AI in advertising raises, including transparency, hypertargeting, and dark patterns. Shek focuses on the question of transparency and whether labeling AI-generated content is needed to enhance consumer trust.
Sidley describes two unique challenges AI presents for advertising: First, it is autonomous, evolving on its own, introducing unpredictability; and second, it is opaque, meaning we often can’t see how an AI system arrived at its output. These properties make AI both powerful and harder to regulate and trust.
As a result, Sidley argues, advertising’s ethical north star must remain consumer trust — the foundation of every successful campaign.
There is no doubt that transparency and trust generally are important to ethical advertising, and in his paper Sidley explores a few important challenges that the use of AI in advertising may present to consumer trust.
Transparency: Should AI-generated ads be labeled as such?
Sidley cites studies showing AI disclosure can boost trust by up to 73% in ad trustworthiness and 96% in company trust (Yahoo/Publicis, 2024). Yet, determining when and how to label is complex:- What if only part of the ad (say, the background image) was AI-made?
- How should advertisers disclose when AI is embedded invisibly in tools?
- Who enforces compliance in a fragmented digital ecosystem?
Sidley suggests a compromise: a simple disclosure such as “Some AI-generated elements may have been used in the creation of this advertisement.”
Hypertargeting: Balancing relevance and privacy
AI can be used to supercharge audience targeting. While consumers want relevant ads, they also resent feeling surveilled. Sidley characterizes this as a privacy conundrum and proposes reforms such as:- True transparency about how targeting works.
- Tools for consumers to set and change privacy preferences.
- Strict limits on targeting based on sensitive characteristics.
- Regular audits of algorithms for bias or discrimination.
Dark Patterns
AI opens new frontiers for deceptive design. Sidley warns that AI may be used to create dark patterns, that is, techniques that go beyond persuasion to manipulation. He hypothesizes the following misuses:- Hyper-personalized emotional manipulation.
- AI-generated fake reviews or scarcity signals.
- “Friendly” chatbots nudging purchases deceptively.
- Reality distortion through subtle visual fakery.
- Addictive mechanisms optimizing engagement features.
Sidley urges the advertising industry to counter these risks through independent algorithmic audits, clear ethical design standards, consumer education, technical countermeasures, regulatory approaches, and whistleblower protections. Without vigilance, he cautions, the abuse of AI could erode digital trust and provoke severe regulatory backlash.
Sidley closes with a philosophical question: Should we even try to regulate AI in advertising? His answer is yes — but carefully. Broad “AI laws” are too blunt. Instead, regulation should focus on specific use cases where deception or harm is likely.
His roadmap for self-regulators and policymakers includes:
- Identifying which AI uses are truly novel and risky.
- Adapting existing codes to cover them.
- Developing “common-sense” conduct standards for new AI tools.
- Updating guidance continuously.
- Monitoring consumer sentiment on AI trust.
- Keeping self-regulation flexible and fast-moving.
The goal is not to hinder innovation but to ensure it serves consumers ethically and transparently.
CISR shares this philosophy. In our recent publication examining the risks of generative AI, among our recommendations for industry leaders is to ensure that AI enhancement does not mislead as to product characteristics, function, or performance. Or more pointedly, it is the responsibility of advertisers to ensure that the ad does not mislead or blur the distinction between what is real and what is imaginary.
Beyond Simple Labeling: A Contextual Approach
Konrad Shek’s think piece provides a nice complement to Sidley’s paper, drilling down on whether it is necessary or helpful to label AI-generated ads.Shek examines the ethical, legal, and policy implications of AI-generated content in advertising, arguing that mandatory labeling is too blunt an instrument to ensure transparency or consumer protection. Instead, Shek calls for a risk-based, context-driven framework that balances innovation, accountability, and consumer trust.
Shek explores how generative AI may differ from traditional tools such as CGI and Photoshop. He concludes that due to gen AI’s accessibility, speed, and ability to produce hyper-realistic synthetic content without human skill, it may pose greater risks than other technologies. Key issues include:
- Convincing realism that blurs fact and fiction.
- Inaccuracy and hallucination, where AI fabricates plausible but false content.
- Potential harms such as misinformation, deepfakes, bias, and privacy violations.
Despite these risks, Shek warns that technology-based, top-down regulation may be too rigid. Instead, he advocates for use-case-specific oversight. Moreover, existing consumer protection and advertising laws already prohibit misleading or deceptive content globally. The question then becomes whether, and if so, what, additional regulation may be needed for AI-created ads.
Public opinion, too, is nuanced. Shek reviews the available evidence regarding public perception of AI-generated ads. Surveys show strong public demand for transparency but persistent skepticism about AI-generated ads:
- ~77% of UK consumers want AI ads labeled, but many trust them less once labeled.
- Younger audiences are more accepting, though overall trust declines with social grievance and misinformation fears.
Despite the superficial appeal of requiring labeling of AI-generated ads, Shek notes that it is probably not the right approach and risks backfiring for several reasons.
First, Shek points to research showing an “implied truth effect,” where when some content is labeled and some is not, consumers tend to distrust the labeled content even though there is no relationship between the accuracy of the content and whether it is labeled. Second, the existence of AI aversion means some consumers distrust AI-labeled content even though, again, both labeled and unlabeled content may be accurate, or not. Third, given that nearly all ads may soon use some AI, universal labeling could become meaningless or ignored.
In contrast to Sidley’s compromise solution of a disclosure that ad content may have been AI-generated, Shek concludes that a more nuanced approach to the issue is required. He proposes a risk-based framework:
- Assess risk based on degree of AI influence, claim verifiability, consumer impact, and medium context.
- Label high-risk uses (e.g., hyper-realistic virtual endorsers) but not minor or obvious AI edits.
- Ensure clarity and proximity of labels; consider visual cues, metadata, and consistency across platforms.
- Promote media literacy and advertiser education to strengthen critical understanding of AI content.
- Iterate continuously based on consumer comprehension and evolving technology.
Shek concludes that AI labeling should not be universal but contextual and proportionate, applied only when deception risk exists. The guiding principle remains that advertising must be legal, honest, and truthful, regardless of the technology used. Effective ways forward, therefore, go beyond simple labeling toward smarter, adaptive, and context-driven approaches that consider the actual risk of consumer deception.
The Road Ahead: Responsible AI Through Collaboration
The timing of the ICAS Think Tank’s thought leadership, as well as Sidley’s and Shek’s recommendations for the path forward, aligns with CISR’s own exploration into AI industry self-regulation.Our AI Working Group, announced just this week, is an expansion of a convening that has been active for the last 15 months under BBB National Programs’ Children’s Advertising Review Unit (CARU), resulting in the development of the Generative AI & Kids Risk Matrix.
The CISR AI Working Group will consider the array of harms that AI chatbots pose when children and teens use them as companions and will consider what safeguards industry should consider to protect against these harms while allowing for the benefits.
AI has transformed advertising with breathtaking speed, reshaping creative design, targeting, and consumer engagement, often outpacing the development of guardrails and governance. The collaborative work of the ICAS Think Tank and CISR’s AI Working Group is essential to ensure innovation moves forward responsibly.
We invite industry leaders, policymakers, and researchers to join us in shaping a future where AI innovation and consumer trust grow together.