"The Anti-AI Crowd", Open Weights and 1500 applications
Three distinct things on my mind this week. Please allow me to process them, by writing here! As always, I share as thoughts and half-baked opinions.
Things that made me happy this week.
Honestly it was a slog of a week, so I am going to reach!
* I won a basket of cat things, at the wonderful 'Cat Walk' in Victoria last weekend. I still have to pick it up but will share a picture next time.
* My strawberry plants are starting to produce fruit!
* I took my little trailer up to the campground tonight, and am going back there once I hit 'publish' here!
I am giving a talk next week at OTESSA 2026 (Open Education) called "Digital Sovereignty with an Open Lens" where I argue that the four freedoms PLUS governance is the framework to consider.
The "Anti AI Crowd"
This past week we saw the rsync drama play out. I will summarize in this way: a well respected maintainer, experimented with Claude and tests for rsync, and some things went wrong for users and a segment of those users were horrible and even cruel about it. I won't weigh in or not on the actual incident- the maintainer has written his own response, and there is no shortage of perspectives.
What I do want to weigh in on how many times I heard some variation of the "Anti AI Crowd" used to describe a group of bad actors. Rather, than specifically about those people's behavior and how unacceptable this behavior is in open source (no matter the topic).
I am not saying - please stop it hurts peoples feelings. I am saying, please stop - because you are trivializing what is a series of complex ideas, counter-movements, and advocacy that is important to everyone to work through. Some distinct areas of tension that I can think of include:
- AI is not a monolith; AI is a complexity of components, that we are still trying to describe and understand in various contexts including 'what is open'. Unlike early days of 'view source', a seat at the table involves a lot more - including compute.
- The narrative about AI and its role in replacing workers, or replacing workers-salaries, or just intensifying work for humans - the narrative wobbles every week, but in each case humans are stressed.
- How and what AI is trained on - with deep significance to creators and the open web.
- Consolidation of power. Ownership and influence of LLMs by some of the most infamous people and companies in history (at this point), is of a great concern to human rights organizations as well as individuals.
- AI use has different implications depending on where and how it's used - and especially if that use is consistent with human values expectations. The problem of AI Alignment was called by Open AI whistleblower Daniel Kokotajlo, one of the greatest challenges facing human kind - and his concern about oppression of dissenting voices.
- The environment - people are protesting data centers, all over the globe - most recently "No AI Vancouver" out my way.
I do get that AI is helping people; I get that there is huge potential for medical discoveries, I do understand that peoples livelihood and progress feels tied to this new technology - but have a lot to lose by dismissing and generalizing dissenting voices - in creating an "us vs them". Everyone could have done better in this case, but the generalization feels super dangerous.
G7 and Open Weights
I have been paying close attention to, and documenting my understanding of "Open Source AI" - and to be honest, my focus is less about the business side of things, than how openness is or is not empowering people in the technology they use every day.
I debunked my own theory that Open Source AI data must be open to be understood (you can read more about that here if you want) . Red Monk did a great job of summarizing the G7 outcomes this week, where it was helpful to learn that paralysis in collaboration between members, has been because the definition of "Open S0urce AI" remained so contested.
Rather than debate, what can or cannot fall under that definition, they have instead chosen to focus on describing AI stacks (my take). For example "Open Source AI with Open Data" articulates the data part, while "Open Source AI" does not include open data. In addition to that "Open Source Weights" (weights with Open Source Licenses).
What the G7 is proposing serves as a recognition that that debate is a lost cause. Instead, for all intents and purposes, the G7 is proposing to deprecate the term open source AI in favor of open weights.
I thought this line was the most interesting, because I really haven't played a lot with open weights until now. This sent me on a bit a wormhole, in trying to understand how weights might be that illusive way to influence model output for open source communities. My next goal is to start remixing existing open weights to see what I can learn (and thus have an opinion on). Will let you know how it goes.
We will see how it goes, as it does seem to have unlocked something - shortly afterwards our prime minster (God help me) announced "AI for All", that should be interesting. What definition is that? 😅
This shared language framework directly unblocks export control negotiations, meaning restrictions on compute flowing to open-model developers could now be coordinated across the US, EU, and Japan rather than applied unilaterally and inconsistently. For founders and practitioners building on open-weights models like Llama or Mistral, this signals that liability rules and access restrictions are moving from theoretical to near-term legislative agenda items across all seven major economies simultaneously. - AI Weekly