Artificial intelligence is making its way into all sectors, and the cosmetics industry is no exception. Skin diagnosis, product recommendation, predictive trends… some even claim that ai is now capable of formulating a cosmetic product from A to Z.
But can a cream or a serum created by ChatGPT, Google Gemini, ERNIE (Baidu), Claude, Perplexity… really be placed on the market in full compliance?
What is the scientific, technical and regulatory reality behind this promise?
Let us take stock of what artificial intelligence can genuinely bring to cosmetic formulation, what it cannot do, and the legal risks associated with poorly controlled use.
AI in cosmetics: genuine innovation or simple marketing tool?
References to artificial intelligence are multiplying in cosmetic brand communication, often through attractive but imprecise claims.
- « Formula developed by AI »
- « Product co-created with artificial intelligence »
- « Patented formulation algorithm »
In the majority of cases, these statements refer to decision support tools, data structuring or ingredient selection, rather than to an autonomous formulation process.
However, cosmetic formulation is based on a strict scientific, experimental and regulatory framework, which cannot be reduced to a simple theoretical assembly of ingredients. In this context, responsible communication consists in clearly positioning artificial intelligence as a formulation support tool, intervening upstream of development, and not as an independent actor guaranteeing product performance, safety or compliance on its own.
What AI can genuinely contribute to cosmetic formulation
Artificial intelligence does not replace the expertise of R&D teams, but acts as a tool to support upstream structuring, particularly during the preparation of the product brief. Even before the first exchanges with a cosmetic manufacturing laboratory, it allows the brand to define a clearer direction, formalise its technical expectations and clarify its needs. It therefore contributes to a shared understanding of the project between the client and research and development teams, fostering smoother, more precise and more efficient dialogue from the initial phases of collaboration.
Its value lies in its ability to cross-reference and structure technical data from supplier sources, as well as from publicly available regulatory and scientific data. This approach helps guide formulation choices from the scoping phase, by contributing to the development of a brief that is consistent with technical and regulatory constraints, by identifying relevant ingredient families in relation to product objectives, and by excluding options that are incompatible with applicable compliance frameworks.
Within this logic, AI helps steer formulation choices from the scoping stage.
It can also support early-stage reflection on key levers such as target cost positioning, formula simplification, integration of more sustainable ingredients, or adaptation to specific requirements such as export markets or certifications, without replacing the development and validation work carried out by R&D teams and regulatory departments.
The regulatory framework for cosmetic development
The development of a cosmetic product, particularly under the european cosmetic regulation, is based on mandatory validations, including stability studies, preservation efficacy testing, skin tolerance assessments, packaging compatibility and the data required for safety assessment.
Responsibility for placing on the market
The use of artificial intelligence solutions at the upstream stage of development does not in any way alter the responsibility associated with placing a cosmetic product on the market. This responsibility remains fully borne by the brand or the entity designated as the responsible person, regardless of the level of assistance provided by digital tools.
The Product Information File (PIF) remains a non-negotiable requirement. It must be compiled in accordance with applicable requirements and include all mandatory elements, in particular a complete qualitative and quantitative formula, stability data, safety assessment elements and compliant labelling. To date, no artificial intelligence solution is capable of generating a complete, traceable and compliant product information file that meets applicable requirements.
What future for AI in cosmetic formulation?
In the medium term, artificial intelligence is not positioned as a substitute for the formulator, but as an assistance tool in support of human expertise. Its role fits within a collaborative approach, aimed at enriching upstream reflection, accelerating certain analysis phases and informing decision-making, without calling into question scientific, regulatory and industrial responsibilities.
In the longer term, AI could contribute to an evolution of formulation practices by facilitating the management of the increasing complexity of product briefs. In this context, certain existing solutions already make it possible, for example, to anticipate stability trends or to identify areas of attention at an earlier stage, in order to support the development phase and decision-making, without replacing regulatory stability studies.
AI, an intelligence accelerator in the service of cosmetics
Artificial intelligence does not redefine the foundations of cosmetic formulation, but it is gradually changing the way it is carried out. By structuring upstream thinking and facilitating the management of ever more numerous constraints, it positions itself as an efficiency lever in support of R&D teams.
Its relevant integration nevertheless relies on a clear understanding of its limitations and on respect for regulatory and responsibility frameworks, which remain unchanged. In cosmetics, the challenge is therefore not to formulate by AI, but to formulate with discernment, using AI as a tool in support of informed and responsible decision-making.
