Open Source Initiative expands its role to AI and machine learning

Over the last 20 years, the OSI has been the guardian of open-source licenses, but now it will also address non-open-source licenses more and AI's role in software development.
Written by Steven Vaughan-Nichols, Senior Contributing Editor
Wright Studio -- Shutterstock

In the beginning, all software was "free software" and "open source." But, as software became commercialized, almost all software became proprietary. In revolt against this, Richard M. Stallman (RMS) took James Gosling's Emacs text editor and relicensed it under GNU Public License (GPL), the first free software license, in 1983.

While the rise of free software fundamentally changed how software was used, not everyone was happy with RMS, and his Free Software Foundation (FSF) took on software licensing. So, in 1998, Christine Peterson, in a meeting with Jon "Maddog" Hall, Larry Augustin, Eric S. Raymond, Bruce Perens, and other free software luminaries, came up with the term Open Source.

The key difference, then and now, between the two is that free software is about the ethics of sharing code, while open source is focused on the practical benefits of sharing code. 

Or, as early OSI leader and currently Project Lead of OS-Climate Michael Tiemann put it, open source was meant to "dump the moralizing and confrontational attitude that had been associated with 'free software'" and instead promote the open-source approach on "pragmatic, business-case grounds."

While those free software versus open-source arguments show no signs of ending, the OSI has bigger fish to fry today.

Also: It's time to stop using C and C++ for new projects, says Microsoft Azure CTO

Over the last few years, open-source licenses have faced challenges from quasi-open-source licenses such as the Business Source License (BSL), Common Clause, and Server Side Public License (SSPL). The arguments for free software versus open source and vice versa are clear. It's much harder for people to get their minds around the difference between a license that's "sort of, kind of, not really" an open-source license and one that's a real open-source license. 

The core distinction between these licenses and the ones that have the OSI's blessing is that any real open-source license must comply with the Open Source Definition (OSD). It all boils down to the idea that with open source, you get the right to use the software code any way you want. Want to run it on a cloud? Go for it. Want to sell the code or a program based on it? You can do that too. How about offering it as a cloud-based service? That's cool too. With these para-open source licenses, these rights are restricted. 

Open Source Initiative (OSI) logo

Making matters even more confusing, many companies that now use these pseudo-open-source licenses started out with real open-source licenses. 

This is a growing trend. OSI Standards and Policy Director Simon Phipps said, "This is another example of a disappointing trend for companies who have retained control of software rights while claiming to offer open-source freedoms pulling the rug when they have gained enough market momentum -- sometimes called a 'rights-ratchet' model. OSI recommends software users pay careful attention to the sustainable presence of open-source freedoms when committing to deploying a project."

Why do they do this? OSI Executive Director Stefano Maffulli explained in an interview at the 2022 Open Source Summit Europe that dropping open-source licenses has "become a trend with companies. They have a similar recognizable pattern. They build up their business and code over five to 10 years. Along the way, they collect contributor license agreements (CLA) that give the company the rights to their code. Then they change the license and take away those rights." So, they use open-source code to grow, but they then blame open source for not working as a business model. Spoiler alert: Open source has never been, nor will it ever be, a business model. It's a development model. 

But that doesn't stop them, Maffulli said, "from blaming open source for leaving money on the table. So, with their backing, venture capitalists demanding they save every single penny and make more money, they abandon their open-source licenses." Now, he understands their pain, Maffulli says, "but how they're solving it is tainting open source."

So, the OSI moving forward is trying to educate businesses and developers alike about what open-source really is, and isn't. 

As part of this, the OSI is putting more emphasis on its ClearlyDefined project. This crowdsourced project was created in 2018 to address this need and boost open-source projects by improving licensing data in software packages. Ironically, this project has been getting support from one-time proprietary companies such as Microsoft, SAP, and Bloomberg. While some newer, onetime open-source companies are backing away from open source, the old-school businesses are realizing open source's value and embracing it more. The OSI is also looking for a full-time community manager for ClearlyDefined

Open source is also no longer just about developers and businesses, far from it. As Deborah Bryant, the OSI's new US Policy Director, explained, "In today's world, complexity increases constantly and policy-making responses to economic and security matters rise as open-source software's role continues to play a critical role in public and societal concerns."

She's right. Only a few weeks after she arrived at her new post, the US Senate Homeland Security and Governmental Affairs Committee Chairman, Democrat Gary Peters, and top-ranking Republican Rob Portman introduced legislation to help secure open-source software. Open-source supporters don't just need a voice in government policy toward open source, they must have one. Government policy decisions will impact our open-source software ecosystem. The OSI, both in the US and in Europe, with Phipps, is addressing this requirement.

Also: Meta AI guru LeCun says most of today's AI approaches will never lead to true intelligence

Moving forward, the OSI will also address the roles that machine learning (ML) and Artificial Intelligence (AI) play in both using and building open-source software. This has become an increasingly urgent issue.

For example, GitHub's AI-base pair programming tool Copilot's ML model relies in part on open-source code. Some open-source developers are understandably upset at this. Maffulli thinks, "Legally, it appears that GitHub is within its rights." That said, don't get "lost in the legal weeds discussing if there is an open-source license issue here or a copyright issue. This would miss the wider point. There is a fairness issue that affects the whole of society, not just open-source developers."

This isn't just a problem with Copilot. Google's DeepMind has its own AI developer system AlphaCode, Salesforce has CodeT5, and there's also the open-source PolyCoder. It's an issue the open-source community must address. 

The OSI will be kick-starting this conversation with four virtual seminars, Deep Dive: AI, in October. These will examine how AI and open source will intersect each other in business, society, the law, and in academia. I highly recommend that you attend if you care about open source and AI. This, and not the eternal battle between free software and open source, is what's going to matter for the next decade of software development.

Related stories:

Editorial standards