In a context of increasing computerisation of scientific and regulatory practices, artificial intelligence (AI) is gradually establishing itself as a promising tool for assessing the safety of cosmetic products. But how far can it go without compromising the rigour and responsibilities of stakeholders?
European cosmetics regulations require that every product placed on the market be safe for human health under normal and reasonably foreseeable conditions of use. This requirement is met by conducting a safety assessment carried out by a qualified person, the safety assessor.
The accurate compilation and analysis of a range of data is necessary in order to produce a safety report that complies with Article 10 and Annex I of Regulation (EC) No.1223/2009.
Given the often prohibitive costs of AI solutions developed in-house or hosted externally, it is tempting to turn to AI tools that are available free of charge. But can these tools, based on non-specific algorithms, really help to automate the conduct of a security assessment without errors?
The benefit/risk balance
The realm of possibilities: when AI accelerates expertise
Many so-called “generative” AIs are available free of charge in their initial versions and can be a real help in saving time and cross-referencing data, especially for assessors who do not have dedicated software. They can also help automate certain steps in a safety report.
The unavoidable risks: what free AI does not tell you
However, these technologies pose a number of challenges that must …
