open source (OSS)
Software whose 'recipe', i.e., source code, is public: anyone can view, use, modify, and redistribute it according to the conditions specified in the license.
In plain terms
Imagine a restaurant that gives you not only the prepared meal but also the exact recipe, saying: cook it at home, modify it, pass it on to your neighbor. Open source is the same in the world of software. Source code is the text written by programmers from which the working program is built.
With closed-source programs, you only receive the finished product and cannot see inside. With open-source software, anyone can check what the program actually does, fix its bugs, or adapt it to their own needs. This is why much of today's internet is built on it: the foundation of Android, most web servers, and the Firefox browser are all open source.
In the AI world, the term is broader and more disputed. For an AI model, not only the program code matters, but also the trained "knowledge" (the weights) and the training data. Many models called "open" only release the weights, not the training details, so it is worth checking exactly what is open in them.
How it works
- The author uploads the program's source code to a public location, usually a code-sharing site such as GitHub.
- You attach a license: this short legal text specifies what can be done with the code and under what conditions (for example, the original author must be credited).
- Anyone can download, run, and review the code, and create their own version; this is called forking.
- Anyone fixing a bug or writing a new feature can submit their proposal to the original project, where maintainers decide whether to integrate it.
- Thus, the work of even thousands of volunteer and corporate developers can stand behind a program, while the software remains available to everyone.
An everyday example
You download LibreOffice to your computer to avoid paying for an office suite, or watch a film with VLC. Both are open source: you use them for free, and anyone can inspect their code. Similarly, many coding agents and AI tools you try at work are built on open foundations.
Who is affected
A common misconception
Limits and risks
- Public code does not automatically guarantee security: just because anyone could review it does not mean someone actually has.
- Many important projects are maintained by a few volunteers, so bug fixes can be slow, or the project may suddenly become orphaned.
- Licenses differ: so-called copyleft licenses require that if you distribute the program built on them, you must also make its code open. Therefore, read the license before business use.
- In AI models, the 'open' label often refers only to the weights, and the license may restrict, for example, commercial use.
Similar, but not the same
- ingyenes szoftver (freeware) – Freeware costs nothing, but you generally cannot see or modify its code, whereas that is exactly the point of open source.
- open-weight model – With an open-weight AI model, you can download the trained parameters, but the training data and training code mostly remain hidden, and the license can also be stricter.
- szabad szoftver – Free software treats user freedom as an ethical issue, while open source emphasizes the practical benefits of collaborative development, although the two communities largely cover the same programs.
Related concepts
szoftver, szoftverlicenc
open-source AI, copyleft
source code, code repository, szabad szoftver, closed source, open-weight model, coding agent
A brief history
The free software movement was launched by Richard Stallman in 1983 with the GNU project, and in 1991 Linus Torvalds released the first version of the Linux kernel. The term "open source" itself was coined in 1998, the same year the Open Source Initiative, which oversees the concept, was established; in 2024, it also published the definition of open-source AI.
Where things stand
By the end of 2026, open source is the foundation of the software industry: most programs contain open-source components, and a significant share of AI tools and agents is also built on it. For AI models, however, what counts as truly open remains a matter of debate.
In the news
- OccupySaaS open-source project uses AI to clone costly servicesNov 4, 2026 · Code
- DeepSeek's open-source Super-Agent prototype positioned as an alternative to Claude CodeNov 2, 2026 · Code
- AI is speeding up open source as its maintainer base agesNov 2, 2026 · Code
- GitHub open-sources Replica prototyping tool; JetBrains shows ThinkRail IDEOct 28, 2026 · Code
- DeepSeek unveils coding "super-agent" and open-sources its architectureOct 27, 2026 · Agent Age
- AI bug reports swamp maintainers as OpenAI offers advice and Firedancer sets a feeOct 27, 2026 · Code
- Only 35% of open source projects have a GenAI policy as AI-generated code floods inOct 26, 2026 · Code
- Claude reportedly proves safety properties in compiled code of 123 open-source librariesOct 26, 2026 · Code
Also known as
open source · open-source software · open-source · open source · open-source · OSS · FOSS · free and open-source software
Sources
This entry was machine-edited and has undergone machine expert review; the “In the news” section updates automatically from Agent Age stories.