As Europe flirts with “reliable proprietary technologies” and India rolls out open source healthcare AI, Amanda Brock is drawing a sharp line between genuine openness, marketing spin, and outright control.
Three months ago I asked Bruce Perens and Amanda Brock if they still thought that the open source AI definition was a bad idea. After that interview, OpenUK and silicon-focused OpenHW, the two organizations that Brock leads, have published an interesting report on AI Openness, and are following closely what I consider one of the most interesting AI landscapes of the upcoming year.
Predictably, Brock’s opinion about the role and place of open source in AI hasn’t changed since my interview. This is evident in the report, whose executive summary starts by saying that “Due to the complex nature of AI, it is more appropriate to focus on the openness of underlying components (software, models, data) than to try to define ‘open source’ AI itself.”
Commenting that statement, Brock reiterated that, when it comes to AI, it’s necessary to “disaggregate, or go to the component level. We are increasingly seeing people asking for software bills of materials or trustable AI bills of materials — a list of components for AI, required for good practice by smart users. This is an example of how the use of AI will inevitably play out as users mature and need to maintain AI deployments at scale.”
As steps in this direction, Brock mentioned the definition of “14 AI elements that could be disaggregated and categorized as closed, partially open, or fully open” by Radboud University, and more recently, the publication of an Open Model Framework and the Open Model, Data, and Weights licence by the Linux Foundation.
The Long Road Ahead
The biggest lesson Brock has learned so far from her AI openness activism is that “while understanding is growing, there is still a lot to be done to ensure that decisions about AI and open source are informed and based on accurate context. We still see open source used by companies that misrepresent their tech as open source when it is partially open and subject to conditions. This continues to cause confusion.”
Another obstacle that Brock sees is one I had hardly considered: digital sovereignty, or more exactly the way it’s too often seen and sold. Brock thinks “there is a huge friction around sovereignty at the moment. It’s a terribly divisive word that I do not enjoy and [consider] a bullshit buzzword at the moment.”
The reason, she explained, is that “smoke and mirrors around sovereignty potentially allow organizations which wish to localize a tech stack and have to rely on open source to effectively create local source, by localizing the open source contributions. But that is a strategy doomed to failure, as few countries or collections of countries can afford to fork or bifurcate the digital stack.”
The truth, Brock said, is that large cloud hyperscalers are all built on open source, and that their infrastructure and tooling has always been the deep tech of open source, more than the user-facing experiences. This is the place where “open source completely dominates and this is the heart of the stack that those wishing to build sovereignty really need to start to understand.”
“Governance through tools not rules,” she added. “That is the holy grail of infrastructure, and without engaging properly with open source it won’t work.”
As a concrete example of what she calls “improper engagement with open source,” Brock mentioned a very strange statement from the Berlin Digital Sovereignty for the EU Summit last week that said (emphasis mine) “open source solutions can play an important role enhancing digital sovereignty, provided they meet high cybersecurity standards and are complemented by reliable proprietary technologies where appropriate.”
That statement, Brock says, neither delivers the general open source the community requested in the Mozilla letter signed by 59 organizations, nor delivers a feasible base for the Eurostack project.
For Brock, such positions show a complete lack of understanding of how both open source and digital infrastructure works in the EU. She doesn’t think the practical implementations that the EU is currently looking at will work, because while they do recognize open source as a cornerstone, they “don’t really understand what that means in terms of being collaborative,” resulting in a de facto isolationism that, in Brock’s opinion (and I find hard to disagree), will ultimately fail.
Look East!
All this makes Brock’s work even more important in India, a country I’ve previously said could teach the EU and others a lot about digital policies.
In February, Brock will be at the India AI Impact Summit — a flagship global gathering hosted by the Government of India under the IndiaAI Mission — where she hopes to host an OpenHQ “AI openness” event.
The AI Openness report mentioned earlier was a step towards that summit, but it wasn’t the only one. Last month, Brock participated in a panel in Bengaluru, where the main topic was access and democratization of AI. At that event, she told me, people were particularly interested in how healthcare can be “an immediate beneficiary of the power of open source in AI, with both the impact of collaboration and localization through data sets and local language training.”
The main proof presented that open source and AI can benefit healthcare, which needs transparency, customizability, and cost efficiency, was the Indian Open Healthcare Network. OHN uses AI to optimize workflows like Scribe (speech-to-text for electronic medical records), and open source makes possible local adaptation, interoperability (using standards like Fast Healthcare Interoperability Resources), and data sovereignty. The architecture and successes of this project, which began in Kerala and is now used across nine Indian states, were covered in detail earlier this year in a video featuring Open Healthcare Network’s Bodhish Thomas.
During a preparatory roundtable for the Bengaluru panel, Brock said, the discussion focused on digital sovereignty and access to quality data. This wasn’t surprising, she said, because the desire to pursue open source approaches to AI in order to gain its benefits without being locked into particular companies or ecosystems is now genuinely global. At the same time, national approaches still differ markedly, especially on questions of sovereignty, ranging from what she called a pragmatic and collaborative stance in the UK to a more theoretical and isolationist one in Europe.
As far as India is concerned, Brock expects that the Indian open source community will have a big role in shaping the AI conversation there — again, thanks to the indirect support of the government. Prime Minister Modi, said Brock, “already stressed that India sees open source as critical to maintaining access and control around an AI that must be open, inclusive, legitimate, and work for collective progress.”
Time will tell how this will pan out, and following Brock’s work is one way to find out. My own final thought? Considering the focus on nationalism of the current Indian government, it will be interesting to see how the country will make open source AI actually inclusive. But the power of all great inventions is that they bring benefits that their inventors and first adopters wouldn’t have imagined, and open source (even more than AI) is really great, isn’t it?
Marco Fioretti is an aspiring polymath and idealist without illusions based in Rome, Italy. Marco met Linux, Free as in Freedom Software, and the Web pre-1.0 back in the ’90s while working as an ASIC/FPGA designer in Italy, Sweden, and Silicon Valley. This led to tech writing, including but not limited to hundreds of Free/Open Source tutorials. Over time, this odd combination of experiences has made Marco think way too much about the intersection of tech, ethics, and common sense, turning him into an independent scholar of “Human/digital studies” who yearns for a world with less, but much better, much more open and much more sensible tech than we have today.
Be First to Comment