Proprietary Software

Every computer program is written by humans in a kind of text form, the so-called source code. This source code is the blueprint of the program. With proprietary software, the manufacturing company keeps this blueprint to itself. Users only get the finished, running version, similar to how in a restaurant you get the dish but not the recipe. In addition, the manufacturer determines by contract what you may and may not do with the program. The word “proprietary” comes from the Latin word for property and means here: the program remains the property of the company, even after you have paid for it.

Why companies lock away their source code

The main reason is money. Developing a large program often costs years of work and many millions of euros. If the source code were public, a competitor could copy it and offer the same product more cheaply. Secrecy is thus the protection of an investment. Microsoft, Adobe, and SAP have built their business models precisely on this.

For customers this has advantages and disadvantages. On the positive side, there is a clearly identifiable party responsible: if something doesn’t work, there is a support line you can call, and often contractual guarantees as well. With a free community project with no company behind it, no one is obligated to fix a problem.

The downside is called dependency, in technical jargon also lock-in. Anyone who runs their entire accounting in a closed program finds it hard to get out again. If the manufacturer raises prices or discontinues the product, you have hardly any choice. You can’t maintain the program yourself, after all, because you don’t own the blueprint.

Licenses, activation, and subscription models

Legally, proprietary software works via a license. This is a usage agreement that you accept when installing. In doing so, you’re not buying the program, but only the right to use it under certain conditions. Typical conditions are: installation on at most one device, no resale, no examination of the inner workings.

Technically, this is secured by several means. Before delivery, the source code is translated into machine code, a numerical form that the computer understands but that humans can barely read anymore. In addition there are product keys, online activation, or an account you have to log in with.

In recent years, the subscription has become established. Instead of paying 600 euros once for a program version, you pay monthly, and the software regularly checks online whether the subscription is still active. A common misconception, by the way, is that proprietary automatically means paid. That’s not true: WhatsApp and the Chrome browser cost nothing, yet their source code is still not freely viewable.

Closed and open AI models

The topic is currently encountered most clearly in artificial intelligence. GPT-5 from OpenAI and Gemini from Google are proprietary. You can use them via a website or an interface, but you get neither the trained model data nor any information about what material they were trained on. If the service fails or the price changes, all products built on top of it are affected.

By contrast, there are models like Llama from Meta or the models from Mistral, which you can download and run on your own computers. An important nuance: these models are often called “open,” but are usually not completely free. The licenses contain restrictions, and the training data generally remains secret. So between fully closed and fully open lies a broad gray zone.

In everyday life, almost everyone uses both worlds simultaneously. Windows, macOS, iOS, Photoshop, and games on Steam are proprietary. The Firefox browser, the Linux operating system, and the core of Android are open. Authorities and schools have been discussing for years how much dependency on individual manufacturers they want to accept. That is precisely why the term regularly appears in business news.