boo-liet shen 👻

ehhh depends on the ROI, investing in a few premium accounts in order to message 20x more accounts with inmail might be a cost they're willing to invest

a still from TRON (1982)

a still from TRON (1982)

a still from TRON (1982)

Still from TRON (1982

I think what's impressive about TRON is that, visually, it's still so much more interesting than any of it's sequels. It's crazy that they tried update this digital world by making it as 'realistically rendered' as possible. It's a complete misunderstanding of the original's charm.

checked out the Tron trending feed, immediately got hit with diapers

really brings me back to the old days of moderating tinder pics

how do I properly order an extra shot of espresso in drinks in french, I keep ending up with like 2-3 drinks at once depending on how much caffeine I'm trying to get 🥀

A screenshot from LinkedIn showing dozens of comments by different people on an unseen post all saying "Wishing You The Best 🎉"

I feel ok leaving the PII here because I'm confident none of these are real people. A very lazy glance at the accounts makes me think it's a LinkedIn spammer making Pakistani accounts

I got excited when I saw so many comments on our job posting and it's like 100+ of these lmao

The reason why is simple: once you "backdoor" an end-to-end-encryption it really isn't encrypted at all anymore -- the end.

As fascism rises worldwide and states increasingly deploy digital surveillance measures, a "backdoor for CSAM" quickly becomes a "backdoor for state surveillance"

3/

This is a perennial debate: child sexual abuse material (CSAM) is now mostly spread through encrypted networks where it escapes detection.

That's a huge problem, but its also a huge problem to backdoor an encrypted chat even if its "just" for CSAM

2/

Germany’s no halts EU plan to combat child sexual abuse content online The European Commission proposed the measures in 2022, but member states cannot reach an agreement, citing privacy and cybersecurity concerns.

Want to call out a really huge (and surprising) win for E2EE.

So since 2022, there's been a bill in the EU that was looking like it was fait accompli to force companies to backdoor encrypted chats to scan for CSAM

It was really bad from a privacy perspective

www.euronews.com/next/2025/10...

1/

I just wonder where he's at now and what he's doing

thinking about this guy I knew in college who got a massive back tattoo of ANTIFASCIST in like 6 inch tall letters across his shoulders

Headed to D.C. next week for the #INHOPE 2025 Summit! I’ll be speaking on a panel about open source tools and methods for detecting OCSEA networks. excited to learn from and collab with everyone!

If you’ll be there, come say hi 👋 let’s talk online safety and what we’re building at @roost.tools

🧵 A network of 100+ BlueSky accounts pretending to be legitimate news syndication, some impersonating BBC News, @EuroNews.com, and @Meduza.io.

Built upon ≥70 Mastodon instances, then bridged to BlueSky using @ap.brid.gy, it feels like a social-media reinvention of the Portal 🇷🇺Kombat concept.

What Zohran Mamdani Knows About Power The thirty-three-year-old socialist is rewriting the rules of New York politics. Can he transform the city as mayor?

eric did an excellent job with this

www.newyorker.com/magazine/202...

thank you!!!!

ily

I want a roast beef sandwich and a milkshake now

Lac Du Saint-Croix as seen from a road above. An unbelievably cerulean lake perfectly reflects the white clouds in the blue sky surrounded by dark green trees

the Verdon Gorge as seen from the Point Sublime. A milky turquoise blue river snakes through a forested and rocky mountain landscape. The trees are various shades of green and yellow as autumn starts to kiss the region

the area around the verdon gorge took my understanding of the word "beautiful" to new heights today

I snapped this pic of Lac Du Saint-Croix after pulling over on the highway. No edits or filters. Just insanity from nature

Behind every swipe: the workers toiling to keep dating apps safe Harsh targets and distressing content have left mental scars on a global workforce of content moderators

www.thebureauinvestigates.com/stories/2023...

Rough read but I always come back to the need for trauma informed tools too

I think it's like a mix of full time mod/t&s staff, outsourced vendor team, and some platforms do yet another version of outsourcing that uses crowdsourcing with randos

don't get me started on dating apps + mod tool access, I always feel like they're not held to the same standards and discourse as social media

There are a few public examples of this too I think, might be part of the tools paper we wrote

also they (like pretty much everyone) use a 3rd party age verification software vendor! one can only be as secure as one's vendors are~

Trust & safety and content moderation are at their purest forms challenges in scaling.

Both the computational cost of scaling in terms of raw volume AND scaling accurate and consistent decisions. The latter is extra hard cause people are unpredictable and human behavior is weird and ever changing

Context Collapse & T&S One of the hardest things about writing on a microblogging service is context collapse, especially when it comes to complex issues — you reply to one person, others take it out of context, and suddenl...

I'll be trying out leaflet.pub moving forward to try to write longer form blogs on Trust and Safety so that I can explain things in a bit more depth.
leaflet.pub/25030fdf-2a8...

Hinge is taking a fairer approach to account banning | TechCrunch Hinge won’t kick you off the app if just one thing on your profile breaks the rules but it does remove the content and prompts you to change it.

This is cool

techcrunch.com/2025/09/30/h...

For as much as Hive has built-in biases (and every model has built-in biases, because they're trained on human data and humans have built-in biases) it is, hands down, one of the more accurate commercially available services out there. This doesn't mean it doesn't suck on a few things!

Absolutely everyone knows that these available services suck at a few things! But they do refine over time (my favorite example of an image Hive in particular used to detect: bsky.app/profile/raha... )

indigo/automod/visual/hiveai_client.go at 3259b215110eb8bffcd26fc340203f828e6dadad · bluesky-social/indigo Go source code for Bluesky's atproto services. Contribute to bluesky-social/indigo development by creating an account on GitHub.

github.com/bluesky-soci... -- even if you can't really fluently read code, you can still see that there's a bunch of specialcasing where they're combining two scores to be like "hey, Hive kinda sucks at this stuff so let's increase the confidence threshold"

Visual Moderation - Overview Introduction V3 (Demo) vs V2 (Enterprise) at a glance V3 API — easiest to try today in the Playground . For developer testing ONLY : 100 requests/day limit. Great UI for testing and proofs-of-concept....

You can see the full list of what the API returns here: docs.thehive.ai/docs/visual-... Sites then take those scores and decide at what threshold to take action. Like, a site that doesn't care about depictions of drug use might completely ignore that category of labels, etc.

The way Bluesky's implementation of Hive's service works is they query Hive's backend automated interface (API) for every uploaded image, and the API returns confidence scores from 0-1 based on "how much the model thinks this thing matches the criteria for this element".

Rudy is the GOAT

and this kind of collective pluralism is what makes his work a real revolution