Should AI Ads Count As Deep Fake in the EU?

1 hour ago 2

Retailers and wholesalers interest that Europe's wide mentation of heavy fakes mightiness unit them to disclose immoderate contented manipulated with AI, adjacent erstwhile the contented was not designed to mislead the public, an manufacture radical missive read.

EuroCommerce, which counts retail giants specified arsenic H&M and Ikea among its members, asked the European Commission to separate AI that "materially misleads" from AI that is simply utilized to edit oregon marque advertisements, for example.

The publically disposable letter, which was dated June 18, comes arsenic Europe prepares to commencement implementing more regulations nether the AI Act this August, including the transparency obligations requiring companies to disclose heavy fakes successful audio and ocular content.

More from WWD

AI Could Boost European Retail By More Than $360 Billion
Pinterest Has New AI Tools For Advertisers
A Fashion Model Is Again Suing Rainbow Shops Over Unsanctioned AI Images

While the radical said it agrees with having greater transparency surrounding deceptive AI content, the deficiency of wide guidelines mightiness tally the hazard of besides saddling "non-deceptive" AI applications with burdensome regulations.

"This includes practices specified arsenic merchandise representation enhancement oregon AI-generated selling materials that are not intended to mislead users," the missive read. "For example, generating an representation of a surviving country to showcase a sofa, oregon enhancing merchandise visuals for presumption purposes."

Commercial contented should not beryllium counted arsenic heavy fake, the radical said, erstwhile it is "not intended to mislead, impersonate individuals, oregon undermine cardinal rights."

The manufacture group, which speaks for the commercialized bloc's retail and wholesale sectors, said the committee should "clearly separate betwixt contented that is materially misleading" and "ordinary commercialized contented that is AI-edited for presentation."

Otherwise, this would obligate "a precise ample stock of AI-assisted content" to comply, diluting the constituent of transparency, not to notation inducing "disclosure fatigue" and undermining user trust, said EuroCommerce.

"Without wide and contextualised thresholds, adjacent insignificant editing could trigger heavy fake obligations, starring to ineligible uncertainty and inconsistent application," it said.

The letter, written by the group's manager wide Christel Delberghe, was addressed to Henna Virkkunen, enforcement vice-president astatine the commission. "This could pb to indiscriminate labelling, reducing the clarity and usefulness of disclosures for consumers, portion besides creating disproportionate compliance burdens for companies," it warned.

Best of WWD