Discussions surrounding what qualifies as “open source” often conclude by referencing the Open Source Initiative (OSI): If software is distributed under a license that the OSI officially recognizes as “open source” according to its established “definition,” it is considered open source.
However, complexities arise when we examine the legal definitions in contrast to the true “spirit” of open source. There are important subtleties in the discussion between open source and proprietary software: Has an “open source company” compromised its project by placing key features behind a commercial paywall? How transparent is the development process? And to what extent can the “community” contribute meaningfully to a project?
For many, being open source extends beyond just the legal permissions to use and modify the code; the surrounding culture, transparency, and governance play pivotal roles.
Most are familiar with Google’s version of Android, which comes pre-installed on a variety of smartphones and tablets, complete with numerous apps and services. In contrast, the underlying Android Open Source Project (AOSP), released under a permissive Apache 2.0 license, is available for anyone to explore, “fork,” and alter for their hardware projects.
By almost any measure, Android is one of the most open-source endeavors. Google has leveraged this fact in response to antitrust claims, pointing out that Amazon has adapted Android for its Fire-branded products. However, this overlooks the separate “anti-fragmentation agreements” that Google established with hardware manufacturers, which limit their ability to utilize forked versions of Android. Unlike projects like Kubernetes, which operate under an independent foundation with diverse corporate and community contributions, Android is primarily governed by Google, with limited transparency regarding its roadmap or community involvement.
“In terms of licensing, Android could be seen as the most thoroughly documented and inherently open ‘thing’ there is,” stated Luis Villa, co-founder and legal counsel at Tidelift, during a panel at the State of Open Con25 in London. “The licenses are exactly what you would expect — but good luck submitting a patch, and good luck figuring out when the next release is.”
This illustrates the core of the discussion: Open source can sometimes be misleading. A lack of genuine autonomy might hinder those wishing to engage actively in a project. This raises concerns about the sustainability of a project, as seen in many open source companies that have transitioned their licenses to safeguard their commercial interests.
“When considering the real accessibility of open source, it’s about more than just the license,” remarked Peter Zaitsev, founder of open source database services company Percona, during the panel. “Governance is crucial because if a single corporation holds control, they can alter the license instantly.”
These thoughts were reiterated in a separate talk by Dotan Horovits, open source advocate at the Cloud Native Computing Foundation (CNCF), who discussed the potential “dark side” of open source. He observed that conflicts often emerge when a vendor-centric project makes changes driven by its business interests and other pressures. “This raises the question: is vendor-owned open source an oxymoron?” he stated. “I’ve been pondering this for several years, and by 2025, this question is more pertinent than ever.”
The AI factor
These discussions are far from being resolved, particularly as open source has become a significant focus in the field of AI.
China’s DeepSeek made headlines with its open source appeal; while the models feature MIT licenses commonly acknowledged as open source, there remain obscurities concerning training data and other elements. This is why teams at Hugging Face are striving to develop an even “more open” version of DeepSeek’s reasoning model.
Meta has similarly promoted its Llama-branded large language models (LLMs) as open source, even though many argue that Llama does not qualify as open source — the models, while admittedly more “open” than others, still impose commercial limitations.
“I have certain reservations about the definition of open source in AI, but it’s evident that Llama’s approach is not open source,” Villa expressed.
Emily Omier, a consultant focusing on open source businesses and host of the Business of Open Source podcast, noted that attempts to “distort” the meaning of “open source” underscore its inherent value.
“This highlights the strength of the open source brand — the effort to manipulate it suggests that people are invested,” Omier stated during the panel.
Much of this interest may stem from regulatory pressures. The EU AI Act includes a specific exemption for “free and open source” AI systems, except for those deemed to carry an “unacceptable risk.” Villa suggests that this aspect partly explains why a company might seek to redefine what “open source” means.
“There are many entities right now who, because of the reputation [of open source] and the regulatory consequences, aim to alter its definition, and that’s concerning,” Villa commented.
Clear parameters
While advocates for stricter criteria that align with the true essence of open source present valid points, establishing clear parameters as defined by a license simplifies the conversation and minimizes subjective interpretations.
What level of community involvement is necessary for a project to genuinely be characterized as “open source”? From both practical and legal viewpoints, confining the definition to the license makes logical sense.
Stefano Maffulli, executive director at OSI, stated that while many organizations and foundations embrace concepts like “open design, community, and development,” these primarily remain philosophical ideas.
“The purpose of definitions is to provide measurable criteria, and concentrating on licensing achieves that,” Maffulli articulated in a statement to TechCrunch. “The global community and industry rely on the Open Source Definition and now the Open Source AI Definition as objective standards they can trust.”
Compiled by Techarena.au.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
