Shifting Left on Human Rights! @torgo@mastodon.social at @FOSSBackstage@floss.social 🙌
Remote
Florian Scholz
@floscholz@front-end.social
mastodon 4.7.3Documentation D(a)emon at @openwebdocs@front-end.social.
Bremen, Germany
140 Followers
128 Following
14 Posts
Joined April 01, 2023
GitHub:
Homepage:
Pronouns:
he/him
Has anyone seen any SBOMs for LLMs?
Like those of you that do „ethical“ or „trustworthy“ AI, do you provide information on sources, training data, hours trained, hardware, … ?
Email sent to my favorite local cinema asking if they would screen "Ghost in the Machine" (2026):
Replying to
@niklasmerz@w3c.social @sovtechfund@mastodon.social same, the cfp has been quite short but I'm very excited about this program.
Looking forward to @FOSSBackstage@floss.social in Berlin on Monday and Tuesday next week.
Phishing attacks can happen to anyone (not just to German government reps).
We wrote about it a while ago: https://developer.mozilla.org/en-US/docs/Web/Security/Attacks/Phishing
Basically, you have two kinds of hints: hints that build trust and hints that bust trust.
We need to design digital spaces in ways that allow more effective detection of trust and distrust. And we need to train people to be able interpret such hints even in moments of stress or distraction.
What does 'Buy European' even mean?
Replying to
@Aepasek I still think about this one, too. Wondering what would be good areas to try this out additionally.
I guess evaluating to what degree we actually need digital solutions in the first place (to escape the tech-solutionism that was mentioned so much as well) is one part.
And then I thought it might be cool to work with existing analog cooperative structures and their identified digital needs and see if they would support/fund cooperative digital offerings (for their own benefit).
Replying to
@dietrich The most popular JS method is Array filter() if I remember correctly :)
See https://github.com/mdn/yari/blob/main/docs/popularities.md for how to get the data.
RE: @openwebdocs@front-end.social
I learned new risk analysis skills when researching and writing about Threat Modeling. It would be a win if a more diverse set of people (not just sec auditors) would take part in TM activities.
Are results of "Origin Trials" private to browser vendors or do they usually get published and feed back into standardization processes?
What kinds of questions are asked in these trials and what is considered a "positive signal" in favor of the proposed API design?
How are alternatives tested? A/B testing? How diverse is the set of people that are part of the trial? Are mostly businesses asked to participate?
Replying to
@patrickbrosset@mas.to got it, so feedback mechanisms are independent of the OT process. Wondering if this could be done better? Collect feedback more systematically in OT phases? (and make findings available publicly)
Replying to
@patrickbrosset@mas.to The feedback various implementors are collecting seems quite opaque to me. Any signal coming out of it doesn't seem all that traceable and so people trust it less.
I guess https://github.com/web-platform-dx/developer-signals is the more open approach to this? Maybe OT participants could give some of their feedback there? (This would of course require a web-feature ID for OT features, though).
Replying to
@tomayac@toot.cafe thanks! I’m not that interested in the launch process details but more in how OTs are evaluated. I was hoping to find more info like this publicly: https://github.com/w3ctag/design-reviews/issues/795#issuecomment-1982796332