Bio-replicated multilevel physical unclonable fluorescent glass labels enabled by artificial intelligence authentication

• Multilevel physical unclonable glass label is constructed based on micro-textures. • Custom-developed barcode software is used to quickly recognize spectral information. • Unclonable micro-texture information is digitized with high encoding capacity. • High recognition accuracy and fast classification speed are realized by AI. • Intelligent authentication strategy delivers a secure, reliable and efficient solution. Physical unclonable functions (PUFs) have become as a groundbreaking solution in modern information security and anti-counterfeiting due to its inherent unclonable nature derived from unique and unrepeatable physical patterns. However, the construction of multilevel PUF labels in glass remains an arduous challenge by means of the high hardness, brittleness and melting temperature of glass. Herein, inspired by the microstructure of lotus leaf surface, physically unclonable fluorescent glass labels with multilevel authentication modes are constructed based on the random micro-textures of leaves and the characteristic luminescence properties of rare earth ions. Macroscopically, the personalized multicolor fluorescent patterns hidden in the labels appear under ultraviolet light radiation, while the uniqueness of the spectral information for the space-selective doped rare earth ions is analyzed and transformed into a barcode that can be quickly recognized by software developed by smartphone as the primary encryption layer. Microscopically, the unclonable micro-texture information is digitized under low-level magnification. At higher magnification, the micro-texture, as an irreproducible pattern with specific shape information, is extracted and trained by artificial intelligence, which enables the high recognition accuracy of 98–100% and fast classification speed of only 10 epochs. The identity authentication strategy driven by custom-developed barcode identification software alongside artificial intelligence can offer a secure, reliable and efficient solution, which is distinguished by a versatile multilevel authentication mechanism adapted to diverse encryption-level requirements in glass.