MIT License
The MIT License is a short legal text that programmers use to release their source code for free use. It permits almost everything – using, modifying, selling – and only requires that the author's name remain in the code.
Whoever writes a program automatically owns the rights to it. Without permission, no one is allowed to copy, modify, or distribute it. This permission is granted through a license: a short text stating what others are allowed to do with the program. The MIT License is one of the best-known of these texts. It is only about 170 words long and permits practically everything, as long as the name of the original author remains in the text. It is named after the US university Massachusetts Institute of Technology, where it originated in the 1980s.
Why companies particularly like this license
In the software world, there are two major families of licenses. One family demands something in return: whoever reuses the code must also disclose their own resulting work. This type is called copyleft, the most well-known example being the GPL. The MIT License belongs to the other family, the permissive licenses. It demands nothing in return.
For companies, this is the decisive point. They can incorporate MIT code into a product that they later sell for a high price. They don’t have to show anyone their own additional code. This is exactly why MIT code ends up in smartphones, cars, and banking software, while stricter licenses are avoided by some legal departments.
The brevity has another practical advantage. Large projects consist of hundreds of foreign building blocks, each with its own license. Conflicting conditions can legally paralyze a project. A license that imposes almost no conditions, on the other hand, fits with almost anything.
What the text actually requires
The license names a list of rights: use, copy, modify, merge, publish, distribute, sublicense, and sell. No money changes hands for this. There is only one real obligation: the license text along with the copyright notice must be included in every copy. This is why, in the settings of many apps, you find a long section with foreign names and years.
The second part of the text is a disclaimer of liability, usually in capital letters. The author provides the code “as is” and guarantees nothing. If your software crashes because of this code, you cannot sue them. You can think of it like a shelf with a sign saying “Take one, no warranty.”
A common misconception: MIT License does not mean that the code belongs to no one. Copyright remains with the author; they merely give away the usage rights. Anyone who removes the copyright notice violates the license and thereby loses the permission. Equally false is the assumption that MIT code also covers patents or trademarks. Nothing about that is stated in the text.
MIT License in AI models and on GitHub
On GitHub, the largest platform for collaboratively developed code, the MIT License is the most frequently chosen license of all. Well-known examples are the web tool React from the Meta corporation and the editor VS Code from Microsoft. Many libraries used to build AI systems are also licensed under it.
The term appeared in the news in connection with AI models. When the Chinese company DeepSeek released its model R1 in early 2025, it did so under the MIT License. This means anyone is allowed to download, adapt, and use it commercially. Such announcements move stock prices, because freely usable models drive down the price of paid competing offerings.
An important distinction should be made here. Many providers call their models “open” but use their own licenses with restrictions, such as user caps or bans on certain applications. The MIT License does not have such clauses. If a report mentions MIT, that’s a strong signal of truly free use.