Product authentication is proving that a product is genuine and not a fake copy. A fake copy of a branded product is called a counterfeit. Authentication is used for sneakers and handbags, and also for medicines, car spare parts and branded shampoos. Counterfeits cost brands money, and a fake medicine or brake pad can harm the person who uses it.
There are three main ways to prove that a product is real. The first is a mark built into the packaging or label that is very hard to copy. Examples are micro-optics, which are tiny optical patterns, and invisible particles that only a special reader can see. The second is a microscope image of the product itself, such as the leather or the stitching, read by a trained model. The page on AI microscope authentication explains this. The third is a unique code on each pack, which a shop or a buyer checks with a phone.
Consider a strip of a common painkiller bought at a pharmacy in Madurai. In India, the packs of many of the best-selling medicine brands must now carry a QR code. The buyer scans it with a phone camera. A web page opens and shows the medicine's name, the manufacturer, the batch number, the manufacturing date and the expiry date. The buyer compares these with what is printed on the strip. If the page shows a different batch, or if the code does not open at all, the pack may be fake.
Codes are strongest when each pack has its own number, recorded at the factory. This is called serialisation and track-and-trace. The number is recorded again at the distributor and at the pharmacy. So if the same number appears in two cities at once, one of the packs is a copy. The page on track and trace explains the idea in general. Some brands use the same codes for marketing too, as described in connected packaging.
Resale marketplaces depend on authentication. A recommerce app that sells an expensive handbag, or a bid-ask resale marketplace that sells sneakers, checks each item before it reaches the buyer.
Brands also search the internet for listings that sell fakes of their products, and ask the marketplace to remove them. This is part of digital risk protection.
AI models can now produce convincing fake images, labels and listings. So authentication is moving to marks and signatures that exist only in the physical product. A camera and a model check them where the product is sold or resold.