Sometime this summer, a ChatGPT user will type “help me find a bold red lip for a dinner party,” and the world’s largest beauty company will be waiting inside the answer. L’Oréal and OpenAI announced a broad commercial partnership on June 17 at the VivaTech 2026 conference in Paris, embedding Maybelline’s makeup virtual try-on tool — powered by L’Oréal’s ModiFace augmented reality platform — directly into ChatGPT conversation threads. The try-on feature has not yet launched, but when it does, it will mark the first time a major beauty brand has built real-time AR makeup simulation into a conversational AI used by more than 900 million people weekly.
L’Oréal Chief Digital and Marketing Officer Asmita Dubey framed the deal around a shift she described as the “11-minute paradox”: research from Google Analytics and eMarketer suggesting that consumers using conversational AI to research beauty products spend considerably longer per session than those doing traditional web searches — sometimes 11 minutes or more. Rather than a frictionless shortcut to purchase, AI chat is creating longer, more exploratory shopping conversations. That behavioral shift is the commercial rationale for the deal. “In the LLM world, what is that experience?” Dubey asked ahead of her VivaTech presentation. “How do we shape that experience?”
L’Oréal’s answer: be structurally embedded in the platform before the question is even asked.
How the Maybelline Try-On Works Inside ChatGPT
The ModiFace integration will operate as a callable tool within ChatGPT’s conversation interface. A user describing the look they want — or uploading a photo — will be able to trigger ModiFace’s face-tracking system, which maps more than 68 facial landmarks including lip edges, eye contours, and jawline geometry. The system then renders a three-dimensional texture of the selected Maybelline product, dynamically lit to match the user’s image. Critically, all image processing runs on-device, meaning the user’s photo is never transmitted to a remote server — a privacy architecture choice that distinguishes ModiFace from cloud-processed alternatives.
ModiFace, acquired by L’Oréal from a University of Toronto spin-off in 2018, already serves what the company reports as nearly 120 million beauty-tech interactions annually across 66 countries and 31 brands. The ChatGPT integration is a distribution expansion — putting that same face-mapping pipeline inside a platform with an audience roughly seven times larger than ModiFace’s existing annual reach.
The technology is not without constraints. Research on AR makeup systems, including platforms built on ModiFace, has documented meaningful accuracy gaps for users with darker skin tones, where foundation color-matching and shade blending degrade under non-studio lighting. The interaction between melanin concentration, subsurface light scattering, and the physics of product pigmentation is harder to model than the visual differences suggest. L’Oréal has said it uses multi-spectral approaches to improve accuracy across skin tones, but no independent public audit of the ChatGPT integration’s performance across the full range of skin types has been published.
The Product Data Feed: What ChatGPT Says When You Ask About Lancôme
Beyond the AR tool, the partnership’s more consequential commercial mechanism involves L’Oréal feeding structured, verified product data directly to OpenAI’s systems for its US portfolio, including Lancôme and Kérastase. This is the beauty-industry equivalent of a Google Shopping catalog feed: rather than relying on whatever scraped, outdated, or third-party retail copy ChatGPT might otherwise surface when a user asks about a serum, L’Oréal is supplying the model with authoritative information at the stock-keeping-unit level.
The structural implication is significant. Smaller beauty brands lack the resources and the relationships to supply the same kind of enriched product signals to ChatGPT’s recommendation layer. As eMarketer’s April 2026 AI Visibility Index — which analyzed more than 5,200 ChatGPT responses across nine beauty and personal care categories — found, ChatGPT already recommends Maybelline in 51% of lip-makeup queries and regularly surfaces CeraVe and La Roche-Posay across skincare categories. L’Oréal’s new data feed is designed to consolidate and extend that organic visibility advantage. Brands that cannot supply verified data to ChatGPT will increasingly be competing against models explicitly structured to surface their rivals.
Dubey put it plainly in Paris: L’Oréal’s ambition is to be “the definitive source of beauty truth in the answer layer.”
A Research Partnership, Not Just a Consumer Play
Away from the storefront, the deal extends into L’Oréal’s scientific operations. The company is using GPT-Rosalind — OpenAI’s purpose-built life sciences reasoning model, designed for evidence synthesis and experimental planning — to map the skin microbiome for its La Roche-Posay brand. The aim is to identify beneficial bacterial strains from the millions of microbes that live on human skin and accelerate the development of next-generation skincare formulations. This is a faster literature-synthesis and hypothesis-generation task than traditional scientific workflows allow, using a model specifically trained to operate on biological and clinical text rather than a general-purpose language model.
OpenAI’s latest image-generation model will also power CreAItech, L’Oréal’s in-house generative AI content platform, which already draws on models from Google, Adobe, Stable Diffusion, and others. OpenAI joins as the newest member of a multi-vendor AI content stack, not a sole supplier. According to L’Oréal’s most recent earnings reporting, CreAItech has reduced the company’s marketing production costs by 40%, with roughly 10,000 marketing staff using it to generate more than 50,000 assets.
L’Oréal’s AI Stack: Seven Years in the Making
It would misread this deal to treat it as a reactive scramble. L’Oréal acquired ModiFace in 2018, its first technology acquisition in a 115-year history. It trained 73,000 employees in generative AI and built L’OréalGPT as an internal tool. It was among the first brands to test Google’s AI Max advertising product, running 800 campaigns across 23 countries. It signed a separate partnership with Nvidia in 2025 to scale GPU capacity and later expanded that relationship into its research division through Alchemi, Nvidia’s chemistry and materials science AI model. OpenAI, in this frame, is the latest — and most publicly visible — layer added to a stack L’Oréal has been assembling with deliberate pace.
Emmanuel Marill, OpenAI’s Managing Director for EMEA, said the company is excited to help L’Oréal accelerate “from accelerating research and innovation to helping employees work in new ways and creating more useful, intuitive experiences for consumers.”
What You Should Know About Ads Inside ChatGPT
Three L’Oréal brands — SkinCeuticals, CeraVe, and Garnier — are participating in OpenAI’s global ChatGPT advertising pilot, which began in January 2026. It is important to distinguish this from the product data feed and the ModiFace AR tool. ChatGPT’s advertising format uses a visually separated, labeled “Sponsored” placement — a tinted box structurally distinct from the model’s organic answers. OpenAI has stated publicly that advertisers cannot pay to influence what ChatGPT says in its conversational responses.
Even so, the co-existence of paid ad placements and L’Oréal-supplied organic discovery signals in the same ChatGPT environment raises a transparency question worth flagging: a consumer who asks ChatGPT for a moisturizer recommendation and encounters L’Oréal products in both layers — organic answers and labeled ads — may not have a clear mental model of how each result was generated. The Federal Trade Commission has signaled ongoing scrutiny of AI advertising under Section 5 of the FTC Act, which prohibits unfair or deceptive practices regardless of medium.
Frequently Asked Questions
When will the Maybelline ChatGPT virtual try-on feature be available?
L’Oréal and OpenAI announced the Maybelline Makeup Virtual Try-On integration at VivaTech 2026 on June 17, 2026, and said it will launch inside ChatGPT during summer 2026. No specific launch date has been announced.
Will ChatGPT beauty recommendations be paid ads or independent results?
The partnership involves two distinct mechanisms. L’Oréal’s verified product data feed is designed to improve organic product discovery — how its brands appear when ChatGPT answers unprompted beauty questions. Separately, SkinCeuticals, CeraVe, and Garnier are participating in a labeled advertising pilot that uses a visually distinct sponsored placement format. OpenAI has stated that paid advertisers cannot influence the content of ChatGPT’s conversational answers.
What is ModiFace and how does the AR try-on technology work?
ModiFace is L’Oréal’s augmented reality and AI beauty platform, acquired from a University of Toronto spin-off in 2018. It uses computer vision to map more than 68 facial landmarks in real time, then renders 3D product textures — lipstick, eyeshadow, foundation — dynamically lit to match the user’s image. All processing runs on the user’s device, so images are never transmitted to a remote server.
Does this deal give L’Oréal an unfair advantage in AI beauty recommendations?
L’Oréal’s decision to supply verified, structured product data directly to OpenAI’s systems gives it a discoverability advantage that smaller beauty brands without similar scale or relationships cannot easily replicate. As more consumers use conversational AI to research and buy beauty products, the ability to supply authoritative catalog data to AI recommendation layers functions as a new form of search optimization — one currently accessible primarily to the industry’s largest players.