Nicola Sirchia

Who is responsible when AI speaks for your brand?

By Nicola Sirchia. Published . 4 minute read.

One-minute version. AI avatar and AI voice, made from my own photo and voice.

The short answer

The brand is responsible, always. Whatever tool wrote the ad, the product text or the chatbot's answer, the customer sees your logo, and the platform's terms usually leave the risk with you. So decide in advance what AI may do on its own, what it may draft for a person to approve, and what only a person does. Then give every AI that speaks for you a named owner.

Why can't you leave it to the vendor?

Because the vendor is not in the room when it goes wrong. An ad platform's AI can rewrite copy and add formats to an ad you already approved. A chatbot can promise something your service team can't deliver. A translation can turn a careful claim into a bold one. In every case, the customer, the regulator and the press will ask the brand, not the supplier.

What may AI do on its own?

I sort every AI task into one of three zones. The line between them is a business decision, so it belongs to marketing and legal, not to the tool.

AI on its own

It saves time and a mistake is cheap to fix.

  • Tag and sort images and files
  • Find gaps in product data
  • Suggest keywords and headlines
  • Summarise customer feedback

AI drafts, a person approves

Most of the value sits here.

  • Product texts in every language
  • Ad variants for each channel
  • Answers to common questions, before they go live
  • Replies to reviews

Only a person

A wrong answer costs trust, money or a fine.

  • Health, safety and environmental claims
  • Prices, guarantees and promises
  • A complaint that has escalated
  • Any decision about an individual customer
Three zones for AI work. Where each task sits is a business decision, agreed with legal.

The middle zone is where most of the value is. In the content factory I built, the AI writes product texts in every language, and corporate and local marketers approve every one before it goes out. That approval step is what lets you scale without losing control.

What changed in the law this year?

Two rules now apply across the EU, and both land on marketing.

What about the assistant that talks to customers?

Treat it like a new member of staff who represents the brand to thousands of people a day. Decide in advance which topics it must not answer, when it hands over to a person, and how it behaves with a rude customer, because that default is usually set by a developer and nobody looks at it again. Then read a sample of real conversations every month.

Who owns it?

The same rule I use for websites: every AI system has a business owner and a technical owner. Keep one list of every AI feature that is switched on, in your own tools and in your vendors' platforms, with its owner and the date it was last checked. The risk is rarely the AI you chose. It is the one that came switched on in a tool you already had.

Where would I start?

  1. List every place where AI writes, changes or says something in the brand's name, including your ad platforms.
  2. Put each task into one of the three zones, together with legal.
  3. Name a business owner for each, and switch off what nobody wants to own.
  4. Check a sample every month: ads as they actually run, texts as they were published, conversations as customers had them.

Sources