Discussion

AI labels tell shoppers a tool was used, not whether an advert made things up

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#10748

AI labels can tell shoppers that an advert used synthetic media without telling them whether its expert, testimonial or product result is real. An AdExchanger essay proposes a more specific, risk-based approach to AI advertising disclosures.

Watch Desk analysis

What happened

Sourin Chakrabarti, who works on AI labelling for advertising at Meta, argues that a generic “AI was used” notice leaves an important question unanswered: what, exactly, did AI change? He says his views are personal and do not represent Meta.

His proposal draws three lines. Routine editing and decorative backgrounds would need no extra warning; simulations that could be mistaken for real evidence should say what they represent and what supports the claim; fabricated endorsements, invented expert opinions and unsupported results should be rejected outright.

Why it matters

A synthetic background and an AI-generated doctor claiming a treatment works are not the same kind of intervention. If both receive the same broad label, shoppers still have to guess where the performance ends and the evidence begins.

Chakrabarti’s argument is that disclosure should explain the role of AI in the claim, not merely announce its presence. That would give advertisers and platforms a more useful test than counting synthetic pixels, while drawing a firm line at fabricated endorsements and results without support.

Our read

This is a useful proposal, not a new rule: the article does not establish that platforms or regulators have adopted this framework. Its strongest idea is also its simplest: tell people what the synthetic content is meant to prove, and show what real evidence backs it. “AI-generated” is a label; it is not a consumer explanation.

What to watch

  • Whether advertising platforms add questions about AI’s role and evidence to their submission processes.
  • Whether regulators or industry groups adopt a risk-based distinction between illustration and fabricated proof.
  • Whether platforms make disclosure records available for meaningful scrutiny.

Discussion spark: Should AI adverts that simulate product results be allowed with a specific warning and evidence, or should platforms ban them because viewers may still mistake simulation for proof?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.