If you work in paid media today, you don’t really get to “opt out” of AI. It’s already embedded in our ad platforms, your analytics stack, and your daily workflows. At a recent Paid Media Council roundtable, it became clear that the question every marketer is wrestling with right now is how much of what we do should be handed over to AI.
We heard perspectives from agency leaders trying to protect brand equity, healthcare marketers navigating HIPAA regulations, and compliance leaders managing disclosure risk and internal AI governance. What emerged wasn’t a debate about whether AI has value in the world of paid media, but a shared concern about brand differentiation, copyright, and compliance exposure.
The throughline was that AI should be an enhancer to your work, not a substitute. The value only shows up when human judgement stays firmly in the loop.
Where AI Adds Real Value
The strongest examples from the roundtable were about AI reducing hours of manual work into something a marketer can use. Two areas stood out:
- Reporting Automation and Insight Narratives
Anyone who’s ever stitched together performance reports across platforms knows the pain of managing multiple logins, exporting reports into spreadsheets, and cross-checking notes, then carving out a clear story for the CMO or client. Now, to create efficiencies, many companies, including Mediate.ly, are building internal AI models designed to aggregate data across platforms, surface insights, and draft narratives that explain what happened in the campaign.
In this scenario, AI is doing the heavy lifting of pulling the insights out of a pile of disconnected data while the team reviews and refines every output. Instead of spending hours assembling charts and copy, marketers focused on analyzing the “why” behind performance shifts and recommending smarter tests. The outcome is higher-quality, more consistent, and higher-performing paid media campaigns.
- Metrics Aggregation Across the Full Journey
Several roundtable participants also described using AI to aggregate and reconcile metrics that previously lived in solos: on-site behavior, channel performance, and down-funnel signals like RFP volume or qualified leads. One solution given was a proprietary, AI-driven system that pulls in website engagement, LinkedIn, and other media placements, then marries that with changes in RFP’s, email replies, and overall client engagement.
While AI is stitching together multi-source data points and providing summaries, marketers are spending more time on what to do with the story the data is telling.
Where Humans Must Stay in the Loop
If you were to ask our roundtable where AI shouldn’t take the lead, the answer would be strategy, audience definition, and creative direction. If you hand core marketing strategy and brand voice over to generic AI tools, you’re not innovating; you're commoditizing yourself.
As one healthcare marketer in our discussion put it, if you want to create effective paid ads, you have to take a step back before AI even enters the picture. Understanding who you’re targeting and what they care about still depends on human context. AI can summarize research data, but it’s up to the professionals to turn those inputs into a shared, cohesive view of the buyer. Otherwise, you risk an overly generalized picture of your audience that fails to resonate.
The same goes for creative and messaging. AI is excellent at giving you a quick rough draft. Still, if you let it automatically generate ads, using the same AI models as most other companies, then you’ll lose your brand’s authenticity. Teams that own the final product, inject unique brand identifiers, and keep creative firmly human-owned are often the ones who stand out and get the most engagement.
In our agency, we are leaning into the development of tools, technology and processes powered by AI to enhance our level of service to our clients, but there will be a human to interpret what is happening and why with campaign performance and optimizations.
Navigating Compliance and Copyright
Relying too heavily on AI can not only damage a brand’s reputation, but it can also open companies up to real legal and compliance damages, especially in regulated industries such as healthcare and finance. A participant working in a public company described how press releases must follow a rigorous verification process; yet some vendors now claim they can “optimize LLM rankings” by tuning how models weigh certain brands. That’s not just ethically murky; it introduces reputational and market-manipulation questions regulators haven’t fully answered yet.
Others flagged emerging regulations, like forthcoming CRTC guidance in Canada on what’s acceptable AI use in advertising, and the reality that many generative models train on copyrighted content in ways that are still being litigated. News organizations already report instances of AI-written stories lifting language from outlets like The New York Times and CNN without attribution, exactly the kind of scenario marketers do not want associated with their brands.
Some practical advice that came out of the conversation includes treating LLM’s as research assistants, not ghostwriters, and maintaining clear data governance models to ensure sensitive data never leaks beyond your company. Ensuring that there’s a human in the loop who’s empowered to say “no” and not just rubber stamp AI-generated content will significantly help reduce compliance, regulatory, and copyright risk.
Building a Human-First AI Framework
If there’s a single principle that sums up our roundtable discussion, it’s that AI is an enhancer, not a replacement for human judgement. It should never be treated as a fully autonomous system. What’s missing in most organizations is a clear, simple framework that tells everyone from interns to CEOs when and where AI should be utilized.
A few operating principles we’ve found useful to apply directly to our client work are:
1. AI is an enhancer, not a driver.
Use AI to accelerate analysis, expedite drafting, and enable experimentation, but humans should still own the brief, the strategy, and have the final say in what goes out the door.
- Draw hard guardrails
Define the “do not delegate” zones: regulated or clinical claims, investor and disclosure content, nuanced brand positioning, and anything where a hallucination by AI could create legal or reputational damage. Be clear about the function of AI within your company and your creative process.
- Fact-checking as a default
Assume AI can be wrong, even when it’s confident. Build in mandatory verification steps for anything that touches external messaging or regulated subject matter.
- Authenticity and distinction matter
People choose to do business with companies and brands that have something unique to offer. An over-reliance on AI runs you the risk of legal issues and simply getting lost in the sameness of other artificially generated ads. Keep the human at the center of the strategic and creative process.
Join us and others for any of our upcoming Paid Media Council Series Roundtables. Come to discuss industry topics that are shaping (disrupting?!) the way we do our work. Check out our roundtables, here: https://luma.com/paid-media-council