The EU AI Act Reaches Beauty in August. Here Is What Brands Actually Have to Do.
From 2 August, national regulators begin enforcing the core of the EU's new AI law. For beauty brands, distributors and retailers that use AI, the obligations are lighter than the headlines suggest, but they are real, and the runway is short.
The Artificial Intelligence Act, Regulation (EU) 2024/1689, has been in force since August 2024 and is phasing in over several years. The date that matters for most of the industry is now close. From 2 August 2026, national market surveillance authorities across the EU begin supervising and enforcing the parts of the law that touch almost every company already using AI in its day-to-day work. For a sector that has quietly folded AI into translation, copywriting, product imagery and customer service, that is a deadline worth understanding before, not after, it lands.
Who this actually applies to
The Act splits the world into providers, who build AI systems, and deployers, who use them. A beauty brand generating product visuals, a distributor running an AI chatbot, a retailer translating packaging copy with a machine tool: each is a deployer. That is the role nearly everyone in beauty now occupies, often without having decided to. The good news is that most beauty uses sit in the law's lighter tiers, limited risk or minimal risk, not the heavy high-risk category built for things like recruitment and credit scoring.
Most beauty uses of AI are low risk under the Act. The duties that apply are proportionate, not punishing.
Two duties that bite in August
The first is AI literacy. Article 4 of the Act requires every provider and deployer to make sure the people using AI on their behalf understand it: what it is good at, where it goes wrong, and how to use it responsibly, at a level that fits each person's role. This has technically applied since February 2025, but from 2 August 2026 national authorities can supervise and enforce it. The EU's Digital Omnibus, the package softening parts of the Act, has eased the wording from guaranteeing a level of literacy to actively supporting it, but, crucially, it did not defer the duty. It still arrives this August.
The second is transparency. Article 50 says people must be told when they are dealing with AI. Chatbots must declare they are machines. AI-generated or AI-altered content, an AI model showing a product result, a scene that did not happen, must be labelled as artificial. A real product photographed with only its background swapped generally does not need a label; an AI-fabricated before-and-after does. Providers of generative tools already on the market before 2 August have a short transitional window, to 2 December 2026, to meet the machine-marking requirements.
What the Act actually asks for
The Act asks for proportionate, documented AI literacy: training that fits your team, and a simple record that it happened. There is no official certificate or accreditation scheme for AI literacy, and none is required. What matters is that the training is done and that you can show it.
What matters is that the training is done, and that you can show it.
The cost of ignoring it
The Act's penalties are tiered. Prohibited practices can draw fines up to EUR 35 million or 7 percent of global turnover. Most other breaches reach up to EUR 15 million or 3 percent. Article 4 carries no standalone fine, but skipping AI literacy entirely is treated as an aggravating factor if something else goes wrong. For a mid-sized beauty business, the realistic risk is less a headline fine than a regulator, a retail partner or an enterprise client asking to see your AI policy and your training record, and finding nothing there.
What to do before August
The sensible, low-cost moves are clear. Map where AI already sits in the business, from the marketing team's image tools to the customer-service chatbot. Give staff a short, role-appropriate session on using AI responsibly, and keep a dated record of who attended. Add a line to your content process so AI-generated or altered visuals are labelled. Write down a one-page AI usage policy. None of this requires a lawyer or a large budget; it requires doing it.
The wider picture is still moving. The Commission only published draft guidance on the transparency rules in May 2026, and the heavier high-risk obligations have been pushed toward 2027. But the literacy and transparency duties are settled, and they are the ones that reach the beauty industry first. August is a start line, not a finish, and the brands that treat it calmly now will not be the ones scrambling later.
The Beauty Insider tracks the rules reshaping the business of beauty. This article is general information, not legal advice; confirm your own obligations against the official EU AI Act text and your national authority.