AI governance has a Marketing problem

Ask any good marketer and they'll tell you a brand promise is like an unwritten contract with your customer. If the promise breaks, trust breaks with it.

Trivago promised customers something simple: find the cheapest room, automatically, every time. In reality, the hotel comparison site's ranking algorithm weighted results toward whichever booking site paid the highest fee per click, instead of the lowest price. Higher-priced offers were displayed as the “Top Deal” in 66.8 per cent of listings. Nobody wrote misleading copy. The algorithm's design did that on its own. In 2022, the Australian Federal Court fined the German company $44.7 million for misleading Australian consumers.¹ While Trivago has no Australian ownership, the Australian Consumer Law reached it anyway, because the law follows the conduct, not the company's home address.

That's the shape of exposure organisations face today, and as an external-facing organism, marketing frequently meets it first, more often than any other function. Marketing doesn't own AI governance any more than one single function owns data privacy. Every function that touches an AI decision carries part of it. But marketing's role is larger than most leaders currently assume. A generative AI tool drafting a customer message. An automation deciding who receives which offer. An AI agent quoting a price or completing a booking with no human checking the result. Each is a marketing decision, made inside a marketing workflow, with marketing's name on the outcome.

So what do we mean by Governance? Here, we mean the decisions, checks, and lines of accountability that determine whether an AI system is used lawfully and in line with what a business has told its customers to expect. Not a rulebook signed off once, but a live question asked every time a new AI system touches a customer-facing decision.