#safetybydesign — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #safetybydesign, aggregated by home.social.
-
When you build tools with abstract and rather arbitrary notions of “safety” baked into them – not addressing a specific threat model – they get considerably weaker
https://alecmuffett.com/article/161817
#SafetyByDesign #censorship #llm #safety #trust
When you build tools with abst... -
When you build tools with abstract and rather arbitrary notions of “safety” baked into them – not addressing a specific threat model – they get considerably weaker
“Gosh, who would ever have thought such a thing?”
https://x.com/banteg/status/2072048614424166679
#censorship #llm #safety #safetyByDesign #trust -
Wie sieht „Jugendschutz auf Augenhöhe“ konkret aus?
1️⃣ Interdisziplinäre Kooperation (Standard!)
2️⃣ Pädagogische Rahmung statt Abstinenz
3️⃣ Plattform-Regulierung mit Biss (#SafetyByDesign)
4️⃣ Echte Partizipation der JugendDas Bremer Modell zeigt bereits, wie diese Vernetzung gelingen kann.
Darüberhinaus: Stimmen aus den überegionalen Medien, eine Zusammenfassung von Buchrezensionen zu Kollecks: Der Kampf in den Köpfen, eine Liste mit Unterrichtsmaterialien sowie wichtige Telefonnummern
-
They are five young women who refused to accept that safety is a privilege.
CindyUp began as a school project about drink spiking.
But the deeper story is about young people turning fear into design, silence into public language, and “be careful” into “why is this still acceptable?”
#CindyUp #YouthInnovation #DrinkSpikingAwareness #GenderSafety #YouthAgency #SocialInnovation #WomenInSTEM #EntrepreneurshipEducation #SafetyByDesign
-
This sets the scene for many more cases like this. While the evidence was striking, the case highlighted how little user safety features in platform design and decision-making. #SafetyByDesign
-
This is very interesting regarding #SafetyByDesign and #Tech policy as a whole. Because what most Internet and Tech Policy people are saying generally amounts to "It is complicated!" "This will have unintended consequences".
-
Been on a bit of a journey reading about Systems Thinking in the last few weeks. Why is that, you may ask? Well: point one, I wanted to understand the arguments for #SafetyByDesign as added to the UK #OnlineSafetyAct which says that services must be "safe by design".
-
Would you like to see a middle-aged man grumble about Italian battery design for roughly 20 minutes?
You would?!
Wow, I've just the thing for you!! -
The IFTAS Connect Library is home to moderation templates, wellness guides, shared vocabularies and more.
Built with and for moderators who care about fairness, evidence, and civil speech.
Browse the library (no sign-in required) - https://connect.iftas.org/library/
-
Mastodon will add quote posts, even though some users don’t want them | The Verge
Remember, kids: Quote-Tweets served no purpose on Twitter other than abuse… or maybe not. Perhaps people who go around demanding “safety by design” might like to reconsider the usability impact.
https://www.theverge.com/news/613376/mastodon-quote-posts-update
-
Reminder: when a platform or tool is lauded for implementing safety features like social-network mass-blocking-lists, those tools can be used to quietly silence *good* people, too…
Quoth rahaeli on Bluesky:
[Pin:] if my posts show up as blocked to you if you repost or reply to them, it’s because you were subscribed to the abusive and vendetta-filled “lolicon artists” or “CSAM supporters” blocklists by @csam.okconfirmed.com [that] were deleted by Bluesky. When a mod list is deleted for being abusive, some remnants of the block stick around. It’s a known bug and there’s no easy way for you to fix it. I’ve reported it to Bluesky and hopefully they’ll resolve it soon. I really, really hope they’ll fix it soon, because I am up to approximately 20 asks about it a day and it is really fucking annoying.
Some would say that Bluesky is to blame for this issue, e.g.:
“leaving big accounts to explain weird bugs/features” is an addressable concern re: bsky I’ve seen raised twice now
…presumably from the perspective that obviously when you subscribe to a blocklist you want Bluesky to track changes to that list dynamically rather than to treat it as a one-off set of blockings…
Which makes sense except that implementing the feature in that way would mean that the blocklist subscriber could never unblock individuals who might be on that list (e.g.) by accident.
Do we want users to entirely outsource their blocklists to random blocklist maintainers?
No, I didn’t think so either.
So: either blocklist-based blocks are held at an individual level and can go stale with respect to the source blocklist, or else they are held at a blocklist-subscription level and require absolute and ongoing trust in the list maintainer, or they turn into an awkward mass of “Alice blocked Bob because of [List] but subsequently manually unblocked Bob, what state is Bob in now?” which would destroy usability and performance on the server end.
This is not a solvable problem, instead it’s a choice and a matter of building a solution and taking the lumps which come with it — as with all “safety by design” propositions.
-
RT @Livingstone_S
#SocialMedia only being required to exclude children from their services because they have refused to embed #SafetyByDesign and #ChildRights. Let's be clear, whether included or excluded from platforms, children are carrying the cost rather than companies. -
Vielleicht eine neue Ära der sozialen Medien: Ein Milliardär möchte in den USA TikTok kaufen. Und die Plattform grundlegend verändern. Denn: Es ist kein Naturgesetz, dass soziale Medien Hass & Spaltung fördern. #SafetyByDesign
-
Lese @hateaid über #SafetyByDesign "Mit ihren #Algorithmen tragen #Plattformen wesentlich zur massenhaften Verbreitung von Gewalt und Desinformation bei – und sie wissen darum: Vor rund zwei Jahren veröffentlichte das Wall Street Journal die sogenannten #FacebookFiles. Die Journalist*innen werteten in mehreren Artikeln interne Dokumente, Chatprotokolle und Präsentationen von Meta aus.
Die Unterlagen sowie Aussagen der Whistleblowerin #FrancesHaugen zeigen: Dem Unternehmen, zu dem unter anderem #Facebook und #Instagram gehören, war die schädliche Wirkung seiner Angebote bewusst. Hass und polarisierende Inhalte würden verstärkt, um eine höhere Aktivität von Nutzer*innen und Werbeeinnahmen zu erzielen, so Haugen. Hingegen würden kaum Maßnahmen ergriffen, um den Schaden einzudämmen."
-
Warum gibt es bei X (vormals Twitter) mehr Hass als bei Mastodon?
Unsere Analyse zeigt, dass weniger digitale Gewalt möglich ist, wenn die Sicherheit der Nutzer*innen bei der Entwicklung der Plattform berücksichtigt wird. Doch auch im Fediverse gibt es noch viel zu verbessern: