Now is the time to define Open Source AI

You’ve seen so many announcements about new releases of AI large language models, foundational models, tooling, services all claiming to be “open” or “Open Source”. But none of them are shared with the same conditions. Look carefully and you’ll find that each adds some restriction somewhere. That’s far from the permissionless, clear message that Open Source Approved Licenses have provided for 25 years. Nick Vidal’s post contains a list of the most popular models. Have a look at their terms of use.

The lack of a shared definition of what Open Source AI means is a problem also for upcoming regulation, like the European AI Act. One of the drafts of the act tries to carve exceptions for “Open Source AI” but nobody knows what that means.

The time has come for the community to define “Open Source AI”, for regulators, but also for developers, lawyers, researchers and end users. We need clarity in order to replicate in AI the success of the Open Source software ecosystem.

For this, we just announced a series of initiatives for the rest of 2023 and into 2024 to define “Open Source AI.” The first is an open call for proposals for the Deep Dive webinar series: we want to hear your thoughts on this topic. We’re all tired of over-hyped conversations, misinformation and confusion. With the Deep Dive webinar series we’d like to focus on identifying the principles we can agree will not further damage society.

The OSI will also start a public, global, multi-stakeholder consultation with the goal to draft a definition. In the next few weeks we’ll share more details.

I hold weekly office hours on Fridays with OSI members: book time if you want to chat about OSI’s activities, if you want to volunteer or have suggestions.

Stefano Maffulli

Executive Director, OSI 

In this month’s Open Source Initiative Newsletter:

• The AI renaissance and why Open Source matters
• Regulatory language cannot be the same for all software
• Why open video is vital for Open Source
• Webinar: ClearlyDefined proceeding towards a clear governance structure
• Another issue with the Cyber Resilience Act: European standards bodies are inaccessible to Open Source projects
• The Cyber Resilience Act introduces uncertainty and risk leaving Open Source projects confused
• Notable open source news

The AI renaissance and why Open Source matters

Sharing knowledge and sharing code has always been a key driver for innovation in Artificial Intelligence. Researchers have gathered together since AI was established as a field to develop and advance novel techniques, from Natural Language Processing to Artificial Neural Networks, from Machine Learning to Deep Learning.

The world of AI is at an important crossroads. There are two paths forward: one where highly regulated proprietary code, models, and datasets are going to prevail, or one where Open Source dominates.

Regulatory language cannot be the same for all software

In reviewing the language and concepts being used in the various draft bills and directives circulating in Brussels at present, it is clear that the experts crafting the language are using their understanding of proprietary software to build the protections they clearly intend for Open Source. This may be the cause of the problems we continue to see as the instruments iterate, especially in the absence of direct consultation.

Why open video is vital for Open Source

The news that the European Commission’s competition directorate (DG COMP) has decided not to conduct a full antitrust investigation into the Alliance for Open Media’s (AOM) licensing policy is to be welcomed, especially for the AV1 CODEC specification (successor to the VP9 CODEC and intended to allow royalty-free, high-quality video streaming). It seems that whispering voices had falsely suggested the reciprocal licensing of standard-essential patents (SEPs) in AOM’s policy is somehow anti-competitive.

Webinar: ClearlyDefined proceeding towards a clear governance structure

The ClearlyDefined project was invited to give an update as part of the OpenChain webinar series. I had the opportunity to share with this global community the project’s mission: to create a global database of licensing metadata for every Open Source software component ever published. This was a great opportunity to introduce our work towards an improved governance structure. ClearlyDefined Community Manager Nick Vidal goes more in depth here.

Another issue with the Cyber Resilience Act: European standards bodies are inaccessible to Open Source projects

There’s a crucial issue here for Open Source. EU policy experts say not to worry about CRA compliance because the EU standards bodies will streamline it. But the European Standardization Organizations (ESO) are corporate-controlled, patent-loving and expensive to engage. Shouldn’t the EU address this if they want Open Source accommodated? Standards & EU Policy Director Simon Phipps explains more in this recent blog post.

The Cyber Resilience Act introduces uncertainty and risk leaving Open Source projects confused

What might happen if the uncertainty persists around who is held responsible under the Cyber Resilience Act (CRA)? The global Open Source community is averse to legal risks and generally lacks access to counsel, so it’s very possible offers of source code will simply be withdrawn rather than seeking to resolve the uncertainty.

OSI in the news

• Open source and sustainability – where’s the gap? Today, one of the biggest problems our society faces is climate change and the rapidly deteriorating environment and there are those that believe that the problem will be fixed (at least in part) by new technologies. 
• Open source community on warpath with Europe: they are not to blame for cyber security problems. In an open letter to the European Commission, a dozen developers from the open source community have said that the cybersecurity law, as currently written, could have a “devastating effect” on their work.
• Defend Open Source from Trolls: Oppose Patent Rule Changes. The U.S. Patent and Trademark Office is inviting public comment on proposals that would eliminate third parties’ ability to help clean up bad patents.