We just wrote an in-depth post about Toxic Content labeling. It presents a new way of defining toxic speech online-- and illustrates the importance of observable features for accurate language model interpretability. Would love to hear how YOU define toxicity, too! blog.zentropi.ai/observations...

I’ve had a very “text-oriented” view of content labeling for a long time, and used the opportunity of our recent launch to lay out some of those ideas in the context of the idea of “toxicity” Interested to know what others think! blog.zentropi.ai/observations...

Content policies are usually private, one-off efforts. You build yours, I build mine, we don't share much about what works or why. This makes sense given products can (and should) set different policies based on their communities, but it leaves us reinventing the wheel. 🧵 1/5

*whispers* you can continue to read me, the pundit who insisted the other pundits were wrong about these conclusions

Go U Bears!

Picture of the East Wing demolition of the White House taken on my flight out of DCA.

I am forgetful about it self-promotion, so dropping a last minute link to note that I’m giving a talk Berkman Klein today. Come check it out if you’re free, or catch the recording later: cyber.harvard.edu/events/autom...

I feel like some of the difference in reactions here also rests on on frequently you have to do a somewhat complex, but very repetitive, task. Taking the time to get these sort of workflows really dialed in is most useful for stuff you do over and over.

Tyranny is brittle.

So, the first part of this is plainly false, both historically and currently. I don’t think it’s a good thing in most cases…but it’s plainly the case that pressuring the people in charge of moderation to either ban (or not ban) people works *All The Time*. It is why people do it!

New Ctrl-Alt-Speech: Moderating is Such Sweet Sorrow with guest host @dwillner.bsky.social who is entirely responsible for bringing up Shakespeare as part of this discussion. (@benwhitelaw.bsky.social will be back next week!) podcast.ctrlaltspeech.com/2315966/epis...

While terrible, this is entirely unsurprising. If you hold serious safety efforts in contempt, this sort of thing is inevitable.

Disney/ABC have a responsibility to refuse to participate in corruption. Kimmel must be reinstated. If Disney/ABC agree to this extortion then perhaps creatives + workers should consider collective action to push back. Same w/buying park + cruise tickets if they bow. People have power. Ask Target

No one who agrees to this is a journalist.

Losing my ever-loving-mind watching the same people who were just clutching their pearls claiming censorship over mean emails from WH staffers to Twitter about COVID misinfo are now HAVING THE FCC CHAIR openly threaten broadcast licenses over a joke about the president AND THE BROADCASTERS CENSOR IT

This is a massive, history making abuse of your power. It will define your legacy and one day you will come to regret punishing free speech and trying to destroy democracy.

This is jawboning. This is what the Freedom Caucus fascists of the Weaponization Committee and their Substack lackeys pretended was happening under some “Biden regime,” but it wasn’t. It was always projection.

We cannot make the headlines blunter people www.theverge.com/policy/77979...

staring straight into the camera and lying. just a despicable person and a poor excuse for a national leader.

We must stand resolutely against political assassination and political violence of all kinds, and just as resolutely against everyone who exploits acts of violence as the pretext or excuse for political repression of political opponents.

We got really positive feedback on the TrustCon workshop we ran on writing good content policies for LLMs...so we're doing it again! If you're interested go sign up here, so we can start to figure out timing: forms.gle/tj7vf7ng8n7R...

The agenda for the Trust and Safety Research Conference is out now. Two days of lightning talks, presentations, networking and more, with @dwillner.bsky.social‬ as keynote. Join us! For the full line-up and times, plus link to register, visit: cyber.fsi.stanford.edu/content/trus...

A sprite of mischief in New York left sunflowers at the Russian consulate.

I mean honestly, people, this is a really good idea. Cut sunflowers make Russian diplomats really really angry. Buy a few sunflowers. Drop them at an embassy/consulate near you. youtu.be/R8tr6Dhn78A?...

The joy we felt when after hearing repeatedly to not expect anything sooner than 18 months when we were told they'd be rolling something out before the end of the year. The absolute testament to the power and ingenuity of the American system of science. Moon landing level stuff, and now erased.

this is actually and truly, huge. That workshop was ridiculous to hear about and I think I saw like a thousand lightbulbs turn on in people's heads at the same time

This looks absolutely amazing and a quick perusal shows it might actually make running a labeler smooth enough that I might be able to do it once we figure out why my brain is melting

OPEN SOURCE SAFETY OPEN SOURCE SAFETY OPEN SOURCE SAFETY OPEN SOURCE SAFETY OPEN SOURCE SAFETY OPEN SOURCE SAFETY OPEN SOURCE SAFETY OPEN SOURCE SAFETY OPEN SOURCE SAFETY OPEN SOURCE SAFETY OPEN SOURCE SAFETY OPEN SOURCE SAFETY OPEN SOURCE SAFETY OPEN SOURCE SAFETY OPEN SOURCE SAFETY

Just tested this on a few that I know Reddit’s existing Hatred & Harassment automation has blindspots for; it built a focused, accurate labeler in under 10 minutes / a dozen examples, & the human readable criteria it built could be dropped into a training manual / erratum / used to build a regex

For 17 years working in trust and safety, I've watched talented people burn out on impossible tasks. The problem isn't the people, it's the systems. Traditional moderation requires months of retraining for every policy change. Only big companies can afford it, and even then it works poorly. 🧵 1/9

Some think trust & safety workers should speak out more. And while we do need voices, we also need quiet work—the kind done with care, without credit. I want you to know: I see you. You are the ones in the arena. And that’s what still gives me hope. anchorchange.substack.com/p/in-the-are...

Hey if you missed it at #TrustCon yesterday, our live Ctrl-Alt-Speech is now on the podcast feed! With special guests @aagh.bsky.social and @ashkhen.bsky.social podcast.ctrlaltspeech.com/2315966/epis...

I think I’d summarize my first #trustcon as the loveliest of people discussing the darkest of topics. It’s truly something else ♥️

5) MOST IMPORTANT TSPA.org @charlottewillner.bsky.social and team continue to just be amazing and bring together an extremely important community to drive change and safety.

#TrustCon reflections: 1) Omg people are doing some great work and a lot to keep us safe in the face of incredible headwinds

Kudos to @charlottewillner.bsky.social for her opening speech at #trustcon calling out how the US government is making the role of trust & safety more dangerous... And how many people couldn't even come to the event because it's not safe to travel to the US.

So excited for #TrustCon this week! We will be publicly unveiling Zentropi, a platform that helps people instantly build their own content labelers. We'll be opening it up for early access and open sourcing the underlying language model we trained for the task so that it is accessible to everyone.

A year ago at #TrustCon, I ran around like a maniac showing people something on my laptop. We'd just gotten CoPE - our policy interpretation model - working. It felt like a huge achievement, validating our ideas about LLM-powered labeling. 🧵 1/7

take this shit, stab it deep into the attachment between low info, low trust voters and donald trump, and twist

the republican party is a modern day Slave Power and the only way forward for this country is to sweep it into the ash heap of history

A concentration camp, but with alligators. Don't look away. The Big Ugly Bill is hideous & we must use it to defeat the MAGA GOP next year. But what he's doing to asylum-seekers & immigrants is where we are becoming an authoritarian regime. the.ink/p/floridas-t...

The people responsible for this are criminals and should be prosecuted the next time we have a real government.

The subway is not scary. Literal children ride it millions of times every day. Fascists and bigots lie about it constantly and have for years because they’re trying to destroy NYC and our subway, and most things wrong with the subway can be traced to their attacks resulting in its underfunding.

the destruction of chattel slavery is one of the great accomplishments in our nation's history and the reason conservatives hate celebrating it is because doing so legitimizes the black counter-narrative of the united states

The guy arresting you shouldn’t look like the guy in a ski mask robbing a liquor store. It’s why we’re advancing legislation to put a stop to secret police-style masked law enforcement.

Why did DEI do this?

ICE just arrested Brad Lander, the NYC Comptroller and one of the leading candidates for Mayor, without grounds. He was conducting routine immigration court work, escorting individuals from hearings. He asked ICE for their warrant - well within his legal rights. This is political intimidation.

A leader of the MN House Democrats is murdered by a Republican pro-life activist. Rather than condemning political murder of his opponents, Elon Musk spreads the conspiracy theory that it was done by Democrats. Journalists don’t need to use any social media site. But they really shouldn’t use X.