Microplastics, which are tiny plastic fragments smaller than 5 millimeters, have been discovered in a wide range of environments, from oceans and rivers to soil and even indoor air. These particles are particularly concerning because most plastics take decades or even centuries to break down. While larger microplastics are easier to detect and classify, studies suggest that only about 3% of all microplastics in the environment are larger than 100 micrometers—roughly the thickness of a human hair. The smaller particles, which are less than 20 micrometers in size, are especially difficult to detect and characterize, making them a major challenge for researchers. A team led by Christel Hassler, a former scientific coordinator at EPFL, and Florian Breider, director of the EPFL Central Environmental Laboratory, has developed a new AI-based technique to identify and classify microplastics as small as 5 micrometers. Using water samples from Lake Geneva and surrounding rivers, the researchers found that concentrations of microplastics between 1 and 100 micrometers may have been underestimated by more than 650 times. Evaluations by EMPA, a Swiss research institute, revealed that six out of 38 samples analyzed exceeded ecotoxicological thresholds, which are used to assess environmental risks. Hassler emphasizes that these findings are a significant warning, though the current situation is not yet alarming. The key question is whether these high concentrations of microplastics pose a risk to both human populations and ecosystems. While the effects of small microplastics have been studied in laboratories, their impact on natural ecosystems is still not fully understood. This requires precise measurements of their environmental concentrations to assess potential harm. Breider explains that microplastics can affect the entire food chain, potentially leading to the accumulation of plastics and other substances they carry. Over 15,000 chemicals are used in the production of plastics, and some of these are banned in Switzerland, including additives that alter the texture of plastics. The situation is further complicated by the fact that microplastics interact with other pollutants already present in aquatic environments. Studying the combined toxicity of these substances is extremely complex, and understanding their impact on entire ecosystems is even more challenging. The new method relies on AI tools that help identify and classify different types of small plastic polymers. Researchers used two algorithms: the first separated microplastics from biological material in the water samples, and the second classified the microplastics into six major polymer types. This approach provides more accurate results than previous methods. The technique is evolving and could eventually detect even smaller particles and new types of polymers. Breider notes that new methods are essential to fully understand the complex issue of microplastics. Researchers now plan to track changes in microplastic concentrations over time in the Geneva Leman watershed and determine if pollution levels are rising. They also hope to expand the method to classify more polymer types and improve detection at the nanoscale. Hassler acknowledges that no single technique will likely cover the full range of microplastic sizes or capture the full complexity of the problem.