Mont Royal Parc, Montreal, August 2026
What a packed last semester! I went back to university for conflict-resolution lectures, started preparing trainings for future journalists, and… did a lot of music!
Working remotely for an organisation in a foreign language is pretty exciting, but also comes with some challenges, such as feeling isolated. I can certainly see how it affects my social abilities over time, and so I've been trying to find ways around it for the past year or so. That means trying to use less screens and finding occasions to collaborate with other humans I can meet, both in my work area and outside this scope.
Dangerzone (Freedom of the Press Foundation)
We've mainly worked on the 0.11.0 release. That means bumping versions, finding why things are broken and iterating. The work has been somewhat frustrating, and I start to recognise that I'm a builder type rather a fixer type: I like to engineer new products, come with ideas and implement them, but when the work is to continue doing much of the same thing, I get bored and see my excitement fade. I've witnessed that myself on multiple projects, and I feel it on Dangerzone as well for the past few months.
I have hope, though, as we're in the process of moving to dangerzone rust which might open some cool possibilities, especially to integrate Dangerzone in other workflows.
We've also been baking a tool to help generate threat modeling, for software architects and security engineers. It's not totally finished yet and so I'll wait a bit before the announce, but that's been fun to work on!
Conflict resolution training
I've started an university course on conflict resolution and mediation, at the Tours University. It's at the intersection of psychology, sociology, communication and management. The course is spread over a year, at a frequency of 3 days every month. I've done the first two introductionary sessions and I like our diverse group and what we've started learning.
It seems we will be exposed to different approaches and I can't wait to start my interships!
Going back to uni means getting an access to the library and god, I forgot how good that feels! I'm very lucky.
I've online notes about that, and I can share them if that's useful. Shoot me an email (or contact me by other means) in case you want the link.
Journalism tech security trainings
I've been asked by Julie to help with lectures on "How to protect sources at the digital era", and that's exciting! I've actually never did this and so we've spent some time to think about the best ways to have a two days training on this.
Should we just reject LLMs and generative AI?
These past months have been filled with thoughts on AI. It's a terrible technology in many ways (for the users, for the planet, for the economy, and for the workers behind it) but it's also a technology that touches my day to day job.
Recent models are now able to automate security research and software engineering, with a quality level that's been hard to believe. And while I would like to leave under a rock, I don't: it's now part of the reality we live in.
I believe we need to face the reality of this technology, both in terms of what's being built around it (data-centers are popping up everywhere, with a land impact that they never had until now), in terms of what it does to our brains (is that lazyness?), and also consider the concentration of power that comes with it.
So… I'm certainly not saying it's perfect: it has many flaws and that's the reason why it's been hard to articulate: I believe we need to at the same time be critic about it while facing what it's really capable of.
And so, what do we do with this? The first move is to reject this technology as a whole. I understand why. I've done it myself and it has some value. I think we should do it where it makes sense, but I also believe it's a way to let some folks down: the ones who already use AI.
I've been recommended this video from Dr Fatima, titled "How to (anti) AI better" where she has a better take than "just don't use AI", that I found enlightening. Here are my take aways:
- Saying to people to "not use AI" makes them feel guilt but doesn't stop them from using it: they just don't disclose it ; She drews a parallel with harm-reduction techniques (think clean sringes for drug addicts) that's been a way to reduce the general drug usage, way more than saying that "drug is bad" ;
- People with more "AI literacy" actually use the technology less, and better, than people with no or low AI literacy.
As a software developper with what I consider "a somewhat strong sense of ethics", I'm still not sure what the right move is related to AI technology. When the smartphone appeared (15 years ago), we were prompt to reject it, for what I think were good reasons at the time.
Fast-forward 15 years later, a good chunk of the population uses these smartphones on a day-to-day basis and it shapes our day to day life, for the good and for the bad. I'm very sad to see how our attention span has shrank, and how our personal data is sold everywhere, and so I can't help but think that the result of this rejection is that big techs like Apple and Google completely trusted the smartphone landscape, to the point that it's currently almost impossible to have a free operating system on our phones.
Don't hear me wrong, here: a lot of energy is now spent trying to have Linux work on phones, thanks to projects like Ubuntu touch, mobian and pinephone, but I believe this train was missed back then, and what we're doing is… still closing the gap.
And so, getting back to LLMs: I can't help but draw a parallel with them: I'm reluctant to work on "ethical AI" because of all the problems mentioned earlier, but the discussion amongst my fellow technologists has failed to materialize so far, mainly because what I've seen hapening is either the rejection as a strategy, or a quick adoption, without ethical guardrails.
I question the validity of these strategies. Being against it is a right move to make, but I think only if we think about what is the other move we should be making at the same time.
I know a lot of technolgists that use LLMs in their day to day lives, no matter what I think about that. This comes with risks, both to their security and to their brains and I think we have to face things as they are rather than as we would like them to be. That means discussing rather than dismissing. That means letting the guilt go away, and talk.
Meantime, this technology has a really quick adoption curve, and just saying "don't use it" does not has the issue we'd like it to have.