The copyright crisis in AI that open source can’t ignore

How open source lost its way with AI and why we need to reclaim its influence.

There was a time, not that long ago, when open source was a force to be reckoned with. Open source communities and foundations had the power, and the moral and ethical imperative, to drive and enforce openness, transparency, collaboration and, last but not least, a respect for copyright and IP.

The latter may seem counterintuitive; after all, isn’t the term “copyleft” a repudiation of copyright itself? It may surprise you to find out that it is not. It is instead a way to use copyright in order to maintain freedoms and access that copyright itself may restrict or disable completely. Open source is based around a clear understanding of copyright, its critical need in various aspects, but just as importantly, as a legal avenue that can be used to promote and empower open source licenses. If copyright itself wasn’t respected, then these licenses would be moot.

Read more: Why open source is critical for the continued advancement of new tech

But in its drive to focus its energies on other issues that many see as outside its scope, we are seeing many entities and people within open source willing to give AI “a pass” on its lackadaisical approach to copyright. We may argue about whether all training data should be available for an AI model/codebase to be called “open source,” but we are ignoring the very real problem with copyright itself being abused in order to promote AI. The mentality is almost “AI is too important to bother about copyright“, which is a dangerous slippery slope. Allowing AI to get away with thumbing its nose regarding copyright allows it to thumb its nose elsewhere, such as open source licensing itself. How can we expect companies to respect and honor the conditions of the Apache License, for example, if they refuse to even honor and respect whole idea of copyright and intellectual property.

And it’s not only copyright itself which is being trampled. Although various “Do No Evil” type licenses are not open source (for obvious reasons), the core tenets of open source envision an environment where Software is a public good; where code and technology is shared and collaborated on, ideally transparently in the public venue. Increasingly however, we are seeing Walled Gardens of AI, under the facade of open source. Even in those cases where an AI project is released and called open source, when its license is clearly NOT, for the most part the “leaders” of open source are quiet, far too often willing to acquiesce in order to appease well funded companies. We should expect more.

We should also applaud those companies that take truly-open AI seriously; those that value the ethical use of AI, and are diligently supporting efforts to form unions and alliances for the common good (such as the AI Alliance). I am humbled to be a small part of that. But they are too few, and we need more.

Read more: Inside the shareware era: Features, bugs, and programming in DOS

The open source community is well placed to be a major player in defining the current and future role of AI. All major technology that has achieved a major foothold in the tech space did so because of influence and availability of open source offerings, and open source philosophy. But we are continuing to miss the boat when it comes to AI.

We also need the open source community itself to provide the guidance and mentoring that the AI eco-system so seriously needs. We need to work together to prevent a situation where such critical technology is a black box, controlled and owned by a few, and they alone judge who is worthy, or not, to have access and control. This is just the type of environment that spurred the Free Software and open source revolution. We should be doing a lot more.

Why aren’t we?

More from We Love Open Source

• Why open source is critical for the continued advancement of new tech
• Inside the shareware era: Features, bugs, and programming in DOS
• Are we coding through a revolution or an evolution?
• Why AI won’t replace developers
• What is prompt engineering?

This article is adapted from “How Open Source Lost Its Way with AI” by Jim Jagielski, and is republished with permission from the author.