Using AI in Marketing Without the Hype

Marketing was the first function in most businesses to adopt AI at scale and the first to discover the limits. There is a version of this that works and a version that quietly damages the thing it was meant to help. There are also disclosure rules that have applied since August 2026 and that a lot of marketing teams have not registered.

What works

The useful applications tend to be the unglamorous ones, where AI accelerates work a competent person was going to do anyway.

  • Getting past the blank page on a draft you then rewrite
  • Producing variations of something that already works, for testing
  • Summarising customer feedback, reviews and support tickets to find patterns
  • Reformatting existing material for a different channel
  • First-pass research to be verified before it goes anywhere near a customer

What does not

Publishing generated content without a subject expert reading it is where most of the damage happens. It is not that the output reads badly – it usually reads fine. It is that it is generic, occasionally wrong, and says nothing a competitor could not have said. In a technical field, customers notice quickly.

The second failure is volume for its own sake. Publishing more pages faster is not a strategy, and search engines have become considerably better at recognising material that exists only to occupy space.

The transparency rules that now apply

Article 50 of the EU AI Act took effect on 2 August 2026 and was not affected by the deferrals in the Digital Omnibus. It matters more to marketing teams than to almost anyone else, because it governs disclosure.

In broad terms it requires that people are told when they are interacting with an AI system rather than a person, unless that is obvious from the context; that AI-generated or manipulated audio, image, video and text content is marked and disclosed as such in the circumstances the Article sets out; and that deep fakes are labelled. Providers of systems generating synthetic content must ensure outputs are marked in a machine-readable format, with a transition until 2 December 2026 for systems already on the EU market before August 2026.

The practical consequences for a marketing function are straightforward. A chatbot on your site should not present itself as a member of staff. Generated images used in campaigns fall within the disclosure regime. And if you are using a tool that produces synthetic content, it is worth confirming with the supplier how it handles the marking requirement rather than assuming it does.

A workable position

Use AI to move faster on work you would have done anyway, keep a person accountable for anything published, and be straightforward about what was generated. That last point is increasingly a commercial advantage as well as a legal requirement – audiences are getting better at spotting generated content, and being caught passing it off is worse than disclosing it.

For the wider picture of which obligations apply to your organisation, see what applies now and what was delayed.

Training

The AI Marketing course covers applied use in marketing rather than general AI theory. The UK equivalent is at AI marketing UK.