Ethical AI and Responsible Marketing Practice
How marketers can use AI powerfully, responsibly and credibly through better judgement, governance and practice
Reflection: Where is AI already influencing your marketing decisions, content or customer experience?
Why this matters now
AI capability is expanding quickly across content, insight, personalisation, workflow and customer interaction
At the same time, trust, transparency and governance are becoming commercial as well as ethical issues
NIST’s Generative AI Profile was released in July 2024 to help organisations manage the distinctive risks of generative AI, while the OECD AI Principles continue to frame trustworthy AI around transparency, robustness and accountability.
Key point: AI value without responsible practice creates avoidable risk
Session aims
By the end of this session, you should be able to:
- Understand the main ethical risks in marketing AI use
- Apply simple frameworks for responsible decision-making
- Recognise bias, opacity and governance gaps
- Use practical safeguards for content, targeting and automation
- Clarify the marketer’s role in trustworthy AI adoption
What ethical AI means in marketing
Ethical AI in marketing means using AI in ways that are fair, transparent, accountable, safe and consistent with human values
The OECD says AI actors should commit to transparency, responsible disclosure and accountability, while UNESCO places human rights, dignity, fairness and human oversight at the centre of AI ethics.
In practice, this means: Being honest about AI use ; Reducing bias and unfairness ; Keeping people accountable ; Using governance, not guesswork; Protecting trust as well as efficiency
A useful model: trustworthy AI
A practical set of trustworthy AI principles includes:
Fairness
; Transparency;
Accountability;
Robustness
; Safety and security
; Human oversight
These themes recur across OECD, NIST, UNESCO and Deloitte’s trustworthy AI framing.
Key point: These are not just technical issues. They are marketing issues too
The marketer’s ethical challenge
Marketing teams often use AI in areas such as:
Content creation;
Audience targeting
; Personalisation
; Insight generation
; Lead scoring;
Chatbots and conversational journeys; Workflow automation
Each creates possible ethical questions around truth, fairness, disclosure, manipulation, exclusion and accountability
Reflection: Which of these uses in your organisation carries the highest trust risk?
Bias is not just a data science problem
Bias can enter through:
Training data
; Prompt design
; Default assumptions
; Audience segmentation choices
; Creative outputs
; Performance optimisation goals
UNESCO’s Recommendation explicitly stresses fairness and non-discrimination, while OECD guidance emphasises inclusiveness, human-centred values and accountability.
Key point: If marketers use biased outputs uncritically, they become part of the problem
Marketing examples of AI bias
Examples include:
Imagery tools that underrepresent certain groups;
Copy outputs that default to stereotypes
; Targeting systems that favour profitable audiences and overlook vulnerable or excluded groups;
Lead-scoring logic that amplifies historic bias
; Recommendation systems that narrow rather than broaden opportunity
WFA’s 2024 responsible AI principles were created specifically to guide brands’ use of generative AI in ways that manage these risks.
Transparency matters
The OECD says people should understand AI-based outcomes and be able to challenge them. The EU AI Act introduces disclosure obligations so people know when they are interacting with AI or exposed to certain AI-generated content.
Marketing relevance: Customers should not be misled about whether content, interaction or recommendation is human or AI-generated when that knowledge matters
“AI actors should commit to transparency and responsible disclosure regarding AI systems.” OECD.
A practical transparency test
Ask:
Would a reasonable person know AI was involved?
Could the output materially mislead without disclosure?
Can we explain where the content or recommendation came from?
Can users question or challenge the outcome if needed?
Key point: Transparency is not about burdening people with jargon. It is about preserving informed trust
Governance is where ethics becomes real
Responsible AI use needs more than good intentions
NIST’s AI RMF and GenAI Profile emphasise governance, mapping, measurement and management across the AI lifecycle.
For marketers, governance may include: Approved tools ; Permitted use cases ; Disclosure rules ; Content review steps ; Escalation routes ; Data source controls ; Logging and traceability
A practical framework: MAP, MEASURE, MANAGE
From NIST, a useful simplification is:
Map the use case and risks;
Measure performance and harms
; Manage through controls, monitoring and improvement;
Marketing application: Map where AI is used in campaigns; Measure quality, bias, accuracy and complaints ; Manage with review points, guardrails and accountability
Key point: Responsible AI needs process, not just policy.
Human oversight still matters
UNESCO stresses the importance of human oversight, and OECD guidance links accountability to the roles people play in the AI lifecycle.
In marketing, human oversight is especially important for:
Sensitive claims
; Brand reputation issues
; High-stakes customer communication
; Personal data use
; Segmentation decisions; Potentially discriminatory outcomes
Key point: The goal is not to remove people. It is to place them where judgement matters most
A second framework: the accountability chain
Ask five questions:
Who chose the tool?
Who approved the use case?
Who checked the output?
Who owns the customer impact?
Who is accountable if it goes wrong?
The OECD explicitly calls for traceability and accountability through the AI system lifecycle.
Reflection: If a customer challenged one of your AI-driven outputs tomorrow, who would answer for it?
Case example: AI-generated creative
A team uses generative AI to create ad visuals and copy at speed
The opportunity is faster production
The ethical risks may include bias, unclear provenance, misleading realism, IP concerns and poor disclosure
In April 2026, ICAS highlighted new global best-practice guidance developed with WFA on transparency in AI-generated marketing creative.
Marketing lesson: Speed is useful, but provenance and disclosure cannot be an afterthought
Case example: AI personalisation
AI can improve relevance, timing and efficiency in personalisation . It can also become intrusive, opaque or unfair if customers do not understand how decisions are being made or if optimisation becomes manipulative
Deloitte’s trustworthy AI framing includes being transparent, fair, responsible, accountable, robust and secure as organisations scale GenAI use.
Key point: Helpful personalisation builds trust. Opaque personalisation can damage it
Responsible marketing practice in everyday use
Good practice includes:
Checking outputs for bias and stereotypes;
Being clear when AI has materially shaped customer-facing content;
Using approved data and tools
; Avoiding invented claims or references;
Testing for fairness across audiences;
Keeping human review for higher-risk use cases;
Documenting decisions and controls
WFA’s principles were created because brands need consistent responsible AI practices in day-to-day marketing work, not just at policy level.
Common mistakes to avoid
- Treating AI outputs as facts
- Assuming speed equals value
- Hiding AI use where disclosure matters
- Using biased prompts or datasets without review
- Deploying tools before governance is in place
- Letting no one clearly own the risk
Key point: Irresponsible AI use usually looks ordinary at first. That is what makes it dangerous
The marketer’s role
Marketers are not passive users of AI
They shape prompts, workflows, targeting, claims, customer journeys and brand trust
That means marketers help decide: What AI is used for ; What customers are shown ; What gets disclosed ; What gets checked ; What is acceptable in practice
Reflection: Are marketers in your organisation using AI, or helping govern it?
A contemporary expert view
Deloitte frames trustworthy AI around qualities including transparency, fairness, accountability, robustness and safety. IBM similarly defines trustworthy AI in terms such as explainability, fairness, transparency, safety and security.
“Trustworthy AI refers to AI systems that are explainable, fair, interpretable, robust, transparent, safe and secure.” IBM.
Practical questions before using AI
Ask:
What is the benefit of using AI here?
What could go wrong for customers, colleagues or the brand?
Is bias a realistic risk in this use case?
Should users be told AI was involved?
Who reviews the output?
How do we correct mistakes?
Can we explain and defend this use?
Key point: If you cannot explain it, you probably should not deploy it
Key takeaways
Ethical AI and responsible marketing practice are not barriers to innovation
They are part of how innovation becomes commercially sustainable, trustworthy and defensible
The strongest marketers will not simply use more AI
They will use it with clearer governance, better judgement, and stronger respect for customers and society.
References
NIST. AI Risk Management Framework.
NIST. Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile.
OECD. AI Principles.
OECD. Recommendation of the Council on Artificial Intelligence.
OECD. What are the OECD Principles on AI?
UNESCO. Recommendation on the Ethics of Artificial Intelligence.
UNESCO. Key facts on the Recommendation on the Ethics of Artificial Intelligence.
European Commission. AI Act regulatory framework and transparency obligations.
WFA. Responsible AI principles.
Deloitte. Trustworthy AI.
IBM. What is AI transparency?
IBM. What is Trustworthy AI?
ICAS. Best-practice guidance on transparency in AI-generated marketing creative.
More webinars like this at Cambridge Marketing College
