- Published on
Identifying fake cosmetics using AI
- Authors
- Name
- DeepHorus
A recent investigation demonstrates the potential of artificial intelligence to help consumers identify counterfeit cosmetics, a pervasive issue in the global marketplace. The research, conducted by a lab specializing in low-cost verification tools, focuses on the significant problem of fake branded products on e-commerce platforms. Studies estimate that two-thirds of branded makeup and skincare products on sites like eBay, TikTok Shop, and Vinted are counterfeit, posing financial and health risks to users. These fakes often contain harmful contaminants like heavy metals and bacteria.
The researcher, who has been purchasing authentic and counterfeit cosmetics for an ongoing study, notes the sophisticated effort fraudsters put into mimicking packaging. To test whether AI could assist consumers, the study specifically examined packages of "Rhode Peptide Lip Tint," a product known to be a frequent target for counterfeiters. The methodology involved taking six detailed photos of three different product packages, capturing both the box and the tube.
The images were then uploaded to Google Gemini, utilizing the Gemini 3.6 Flash model with "Thinking" enabled. The AI was prompted with a direct question about the product's authenticity and then asked to identify further red flags. This process simulated a consumer using an AI tool for a quick verification before a purchase.
The results varied by package. For a confirmed counterfeit product purchased from eBay for $5, Gemini correctly identified it, citing specific typographical errors like "Diisosteary! Malate" in the ingredients list. This demonstrates the model's capability to spot minute inconsistencies that evade casual human inspection. The analysis highlights how AI can flag telltale signs of fraud, such as discrepancies in printing and packaging details.
This application of generative AI offers a promising, accessible layer of protection for consumers. It empowers individuals to make more informed decisions without requiring specialized expertise. The research underscores a practical use case for tools like ChatGPT, Gemini, and Claude beyond general knowledge, turning them into potential allies against online scams.
Looking ahead, the success of this preliminary test suggests a pathway for broader consumer tools. Integrating such AI-driven verification could become a standard practice on shopping platforms or through dedicated apps. While not yet infallible, the technology provides a valuable first line of defense against the growing tide of sophisticated counterfeit goods in the beauty industry.
