AI Isn't Killing Accountability in Marketing — It's Exposing Who Never Had It

A bookstore's holiday ad campaign went live with garbled copy and a swapped product image — and no one on the marketing team had touched those changes. Meta's advertising AI had altered the already-approved creative after it was published. The team only found out through complaints from photographers who'd heard their work being called "AI slop." The incident is a vivid illustration of a problem that's new to marketing but rapidly intensifying.

According to Gay Hanson, VP of client engagement at Validity, the issue wasn't the technology — it was the process: "The tool made an unrequested, unflagged change to approved materials, but there was no control step in the process to check the live creative against the approved original after publication." Her proposed fix is simple: lock creative against automated changes once it's approved, or run a scheduled check within 24 hours of going live that compares the live materials against the approved files.

Hanson breaks a typical campaign into three stages: strategy, creation, and approval/delivery. AI has taken over the middle stage first — generating headline variants, copy variants, audience segments and images — while strategy and final sign-off were meant to stay in human hands. In practice, Hanson notes, it's the approval stage that most often falls through the cracks. At large companies the approval chain is longer, and an error can pass through several sign-offs without anyone treating it as fully their own. At small companies, one person runs the whole campaign — which looks like clear accountability, but because that person is juggling strategy, execution and oversight at once, they don't catch what the AI actually changed.

Validity's "2026 State of Email" report — a survey of 502 marketing professionals across the US, UK, Australia and New Zealand — backs this up with numbers. 35% of companies now prioritize AI and machine-learning skills when hiring future email marketers, and 27% prioritize marketing automation and process design. Knowledge of compliance and data privacy, meanwhile, comes up in just 15% of responses, while design and HTML/CSS template-building skills — once considered a company's "single top priority" — have fallen to a mere 14%.

Hanson also lays out the economic logic behind those numbers: lifecycle automations generate 41% of total email revenue even though they account for only about 5% of emails sent — which makes automation an easy investment to justify in a budget meeting. But no one gets applause in a budget meeting for "the lawsuit that didn't happen" — and that invisible payoff is exactly what investment in compliance delivers.

Hanson recommends three concrete steps: audit existing processes and assess what risks AI could introduce; verify that the legal basis for consent obtained from customers actually covers new ways AI is being used; and update privacy policy to match actual practice. In her view, every AI agent capable of directly touching a customer needs a specific named human owner — someone who knows exactly what the agent is allowed to do, what data it can use, and when the process should stop and wait for human intervention. Otherwise, she says, a team doesn't have accountability — it just has hope.