Hey there 👋

You’ve probably ignored a GPS instruction that felt wrong. Your gut says go straight, but the robotic voice insists on a sharp left into a sketchy alley, and for a moment, you have to decide who to trust.

A lot of us are having that feeling with AI in marketing right now.

We’re handing over huge responsibilities: our ad budgets, our customer conversations, our brand’s voice to systems we don’t fully comprehend, and while the upside is huge, the potential for things to go sideways is, frankly, a bit terrifying.

This issue is about how we build the systems and principles that let us use these incredible tools without losing our customers’ trust or our own peace of mind.

Let's go! 🚀

Avoiding AI Pitfalls in 2026: Lessons Learned from Top 2025 Incidents. In 2025, many organizations discovered that AI incidents slipped into everyday operations and created problems where few were looking. A customer chatbot gave confident but incorrect advice. A facial recognition tool led to wrongful arrests. A deepfake encouraged people to invest. Yikes! [Source: ISACA]

Responsible AI Use In Marketing: Navigating Ethics And Consumer Trust. Marketing is fundamentally about trust between brands and consumers. AI’s power to analyze vast datasets and automate decisions can enhance customer experiences, but it also raises critical ethical questions. For example, how do we ensure that AI-driven marketing respects privacy, avoids bias, and remains transparent? [Source: Forbes]

Defensive SEO: How to protect your brand narrative in AI search. AI search summarizes your brand before users visit your site. Defensive SEO helps you monitor and shape your brand narrative in AI search. [Source: Search Engine Land]

There's no shortage of ways AI can go wrong in marketing, but most of the real damage comes from five specific failure modes.

Here's what they are and what to do about each one.

Algorithmic bias occurs when AI systems produce unfair outcomes because the data they were trained on reflects existing inequalities.

In marketing, this shows up in ways that can be subtle but damaging, such as ads that are served disproportionately to certain demographics and audience targeting that excludes people based on protected characteristics, not through deliberate intent, but through the quiet logic of a model that nobody properly audited.

Auditing your training data before deployment, using tools like IBM's AI Fairness 360 or TensorFlow's Fairness Indicators to check for skewed outputs, and building regular human review into your workflow, not as a one-off exercise, but as an ongoing practice, is highly recommended.

Hallucinations occur when AI systems generate confident answers that are simply wrong. The model produces plausible-sounding content that is factually incorrect, uses invented statistics, fabricates quotes, and invents product features, all with complete confidence.

A hallucinated claim in an ad, a made-up statistic in a case study, and a fabricated customer testimonial are potential legal liabilities. ISACA's review of 2025 AI incidents was blunt about it: "Hallucinations are not quirks. They are safety risks."

The practical response is to treat every piece of AI-generated content as a first draft that requires human verification.

Salesforce’s approach is instructive: the company limits model outputs through containment rules and introduces “mindful friction,” intentional pauses in workflows that require human review before AI-generated content is used.

Build the assumption that your AI will sometimes be confidently wrong into every process that touches customer-facing content. Log outputs, version-control your prompts, and create clear escalation paths. 

AI systems run on data. With personalized marketing, the more customer data you rely on, the greater your exposure to privacy risk.

The consent gap is the distance between what consumers think you're doing with their data and what you're actually doing.

Most people have no idea that their browsing behavior, purchase history, and social media activity are being fed into models that predict what they'll buy next, when they're most likely to click, and how much they're willing to pay. They've technically agreed to it in a T&C document that nobody reads, but informed consent and legal consent aren’t the same thing.

The practical implication is straightforward: conduct privacy impact assessments on your AI systems, not just your data collection practices. Understand where personal information enters your AI pipeline, where it's stored, and how it's used. And communicate that clearly to your customers, not in legalese buried in a footer, but in plain language that makes sense.

In 2025, manipulated AI-generated media featuring Mark Carney and fake ads featuring Canadian Prime Minister Mark Carney circulated online, promoting fraudulent investment and trading platforms.

The AI-generated audio and video were convincing enough to deceive viewers, particularly older consumers, into losing their savings. That's an extreme example, but it illustrates a problem that marketers are already encountering in less dramatic forms.

AI-generated images, voices, video, and copy are now indistinguishable from human-created content for most consumers. When a brand uses synthetic content without disclosure, it speaks volumes about its relationship with its audience.

The answer is to be honest about when you're using synthetic AI images or videos, and when transparency about AI use is done right, it builds trust rather than erodes it.

The rule of thumb is simple: if a consumer feels deceived by finding out that AI was involved, you should have told them upfront.

The most seductive trap in AI-powered marketing is the promise of full automation. Set it up, let it run, and watch the results roll in. The problem is that algorithms optimize for what they're told to optimize for, and they're very good at it.

Unilever’s AI governance emphasizes that decisions with significant life impact should involve human oversight. AI incidents have shown that the biggest failures aren’t technical; they’re organizational with weak governance, absent oversight, and a culture of "the algorithm knows best" that created the conditions for harm.

The technology was doing exactly what it was designed to do, but the problem was that nobody had thought carefully enough about what it should be designed to do.

Navigating this ethical minefield requires more than just good intentions; it needs a clear, actionable framework. Here are the three pillars that can guide your strategy.

This is about being clear and upfront about how you’re using AI in ways that affect the customer experience.

AI bias isn’t a hypothetical problem. It’s a real and present danger. AI systems learn from data, and if that data reflects existing societal biases, the AI will amplify them at a massive scale. In marketing, this can lead to discriminatory ad targeting, stereotypical content, and the exclusion of entire customer segments.

AI is a tool, and like any tool, it can be misused. Ultimately, a human must be responsible for the outcome; you can’t blame the algorithm when things go wrong.

Before deploying another AI tool, organizations should clearly define their principles for ethical AI use. This includes outlining what they are willing to do, what they won’t do, and what oversight mechanisms will govern AI use. The policy doesn’t need to be lengthy. A clear, honest one-page document is often more effective than a comprehensive policy that nobody reads.

The quality of your AI outputs is directly determined by the quality of your training data. Your historical data will almost certainly reflect data biases that your model will reproduce. Ensure you audit the data and clean it continuously as it evolves.

You need a genuine quality control mechanism. The marketers who are using AI most effectively are the ones who've figured out exactly where human judgment adds the most value and protected those moments.

When AI is involved, tell them what data you're using, how to opt out, and do so in plain language, and make it easy to find. Customers who understand how you use AI and trust that you're doing it responsibly are more loyal.

Models trained on last year's data may not perform as well on this year's data, and outputs that were fair in one context may be biased in another. Regular audits using both automated tools and human review are not optional extras.

This week's prompt is designed to help you audit your current AI marketing practices against an ethical framework. Use this with Claude, ChatGPT, or any capable LLM. 

Prompt: "I'm the head of marketing at [company name], reviewing our AI-powered marketing practices for ethical risks. Our current AI use includes [list your AI tools and use cases]. 

Please help me identify:

Keep the language plain and practical. This needs to work for a marketing team, not a compliance department.”

These are the most popular AI tools we use at Rise Up Media. If you're not using them already, they're worth a look.

Full disclosure: some links above are affiliate links. If you sign up, we’ll earn a small commission at no extra cost to you.

When you spend as much time on AI Twitter (or X) as I do, you do get a laugh a lot. 😁

The ethics conversation is inseparable from the governance conversation, and governance is where most marketing teams are still figuring things out. 

The tools and technologies will continue to evolve at a dizzying pace, but the fundamentals of trust, honesty, fairness, and accountability are timeless.

Brands need to build their AI strategy on a solid foundation in order to prove through their actions that they’re using AI to create better, more valuable experiences for their customers, not just to cut costs or chase clicks.

It’s a minefield out there, but with the right map and a clear moral compass, it’s one we can all navigate safely.

Until next time, keep exploring the horizon. 🌅

Alex Lielacher

P.S. If you want your brand to show up in Google AI Mode, ChatGPT, and Perplexity, reach out to my agency, Rise Up Media. That's what we do.