It’s vibes like this that want more often for this site.

Fairly certain based on our enforcements that teleguard was used heavily in some child focused communities. Turns out none of that was really encrypted. Oops.

2026 question: If I gave a friend a bunch of RAM I didn’t use for a gaming computer (long story - double order back in 2024), do I still need to buy him a wedding gift?

Australia’s Social Media Minimum Age law has been in effect for three months. The eSafety Commissioner’s first report shows a rough rollout — platforms are missing key obligations and early harm indicators haven’t moved. @markscott.bsky.social and I report: www.techpolicy.press/early-lesson...

in our defense, the buttons are like, right next to each other

Jerry today.

✅ @jcsalterego.bsky.social has been verified by @bsky.app.

if ai makes it easier for devs to create apps, it can do the same for non devs if ai makes it easier for devs to create feeds, it can do the same for non devs if ai makes it easier for devs to create community spaces, it can do the same for non devs

whoops pressed the wrong button

in the next update: we will be unverifying @jcsalterego.bsky.social

Has been on my calendar for months.

Fascinating. Didn’t consider this harm scenario but I also haven’t worked deeply with community notes.

Every time I make the (correct) statement that social media use is not universally bad for teenagers, or state that the best consensus is that it's beneficial-to-neutral for many and any negative effects are a small subset of teenagers, people demand citations (and then refuse to believe them, sigh)

Lessons for community labelers: Skywatch is a part of Bluesky's composable moderation, as a community-run labeler that people can subscribe to labels are a type of metadata about account/content. They exist on their own and as part of API responses or directly embedded in records themselves.

Looking forward to some big changes to verification coming this week.

Required reading for everyone following @kissane.myatproto.social’s awesome #AtmosphereConf keynote.

Exclusive: The Albanese government offered to pay pro-teen social media ban group 36 Months to promote policy. Weeks before the ban began, 36 Months answered a selective departmental invitation to pitch "communications services" to promote the ban, but was unsuccessful. www.crikey.com.au/20...

Just finished up three days of travel and talks. First on regulation, then on trust and safety, and today’s was on tooling for countering extremism. Along the way I tell a number of stories and feel that it’s a good way to translate the madness of T&S at a startup. Here’s one from two years ago.

✍️ Read our full response to Meta’s announcement that it will use more advanced AI technologies in its content moderation: www.oversightboard.com/news/metas-l...

@aaron.bsky.team shouting out how great using Osprey is for investigating! Go @roost.tools @julietshen.bsky.social #opensource #tssummit #trustandsafety

Ahahaha. A speaker just called the T&S and the compliance teams the ‘no teams’. Def a challenge to support and enable the product with safety and legal compliance also in mind.

Next up - folks from AI companies talking about how to integrate child safety.

Next up - folks from AI companies talking about how to integrate child safety.

Let’s start a new thread on AI agents (talk I’m listening to). Systems that can make plans to execute complex tasks. You need the core model, the guard rails for the model (agentic harness), tools that the model can use (ie your email), and the environment it runs on.

Attending my last conference for the next few months, but will thread some interesting ideas that emerge this week that are related to Trust and Safety or Regulatory Compliance. I’m sure this will destroy the thread. No speakers identified cause the Chatham house rule still applies.

Watching in horror as people in the back row enter by the front staircase of the plane and people in the middle/front enter by the back staircase. Some folks do not understand the assignment.

By 2024, the stated goal was a thousand targeting decisions in an hour. That is 3.6 seconds per decision, or from the individual “targeteer’s” perspective, one decision every 72 seconds.

Week 2 at Bluesky Happy to report that week 2 has been just as energizing as week 1. I’m still meeting lots of new people and learning the ropes. In addition to spending time... https://toni.org/2026/03/20/week-2-at-bluesky/

lol.

If Kagi Translate powered inline translations on Bluesky, this output could become officially supported. Just saying 😎

v1.119 is live! Bot labels are here. If you run an automated account, you can now mark it as such in your account settings. The label will show up on your profile and posts so people know what's a bot and what's not.

Honestly this panel alone made my day, so I'm really happy to be learning something new (for a threats nerd like myself) on LLM poisoning, and while it's pretty disconnected cause it's different speakers, gave me much much food for thought.

The panel I've been excited for in Brussels is on - relating to AI and Foreign Interference. It's Chatham House - so I'll do some paraphrasing of interesting points, but won't credit any of the speakers individually.

We hear a lot of, "I really just need..." when we ask developers about their platform safety challenges. While some are platform-specific (so won't be shared w/o permission), so many are universal requests. We started a "safety tools wishlist" for more ideas and upvotes: github.com/orgs/roostor...

Caught up with a friend that I worked with after the conference. He says, "I grabbed some beer and some cheese". In Brussels it seems you can just order half a loaf of bread and a block of cheese as a snack. I see this is how the Europeans live, and possibly why we can't have nice things.

One of my fav threads of all time. Kudos to the account that recreated some of the better posts with alt text bsky.app/profile/juan...

The 4th EEAS Threat Report on FIMI is out and shows the scale of information warfare targeting the EU and its partners in 2025: - 540 incidents - 10,500 channels - 90 targeted countries - 35% of detected incidents linked to Russia and China. 1/3

I've been working in or adjacent to disinformation for far too long. Watched non-state actors adopt the tools of the internet rapidly, then state actors pick up the same playbook. From ISIS to Russia and China — two sides of the same coin. (1/8)

Listening to the Prime Minister of Montenegro today in Brussels on their efforts against foreign manipulation. Look forward to later panels on LLM poisoning and might even live post if that's of interest to folks.

This may be rather late, but I am finally watching severance. Look forward to my first take of Sinners in 2032.

Heading out on a work trip to find out that my pants have a hole in them. I don’t look forward to trying to find another set of decent jeans as a non-youth. It’s always too skinny/baggy/ and the worst is when all the irish men decided to show their ankles off to the world.

Like @toni.bsky.team mentioned, we're only 42 team members. We also happened to hit 43m users last week. Figures like >1m users per person speak to the caliber of the brains at Bluesky. Come be another brain building the future of social: bsky.social/about/join

I appreciate the sheer level of effort that goes into these threads for a museum collection.

I'm thrilled to announce that I'll be joining Bluesky as interim CEO. I deeply believe in what this team has built and the open social web they're fighting for. More here: toni.org/2026/03/09/c...

Some personal news: I’m transitioning from CEO to a new role as Bluesky’s Chief Innovation Officer! I’m excited to welcome @toni.bsky.team as our interim CEO. More here: bsky.social/about/blog/0...

We are aware of recent reports regarding targeted phishing attacks that have resulted in account takeovers of some Signal users, including government officials and journalists. We take this very seriously. 1/7

A response to someone on adult labels, but explaining some of the work ahead on this front for 2026.

Played Spirit Island for the first time last night. It’s definitely not for the faint of heart. 4 hours of playing later and I think I may understand the basics. It’s the most complicated board game I’ve ever played.

Pondering whether or not to do this as an April Fool's joke this year.

Made it to Friday but at what cost