We take community notes for granted now but I can’t emphasize enough how hard it is to pull off an entirely new social mechanic - build trust, acceptance, etc. So many ways it could have gone wrong, like social features often do. Kudos to the team.

The Best of TED 2026 IMHO Community Notes is one of X's greatest advances, a breakthrough in social engineering with a distributed, scalable truth-seeking algorithm. My photo from the front row here (showing a peek into what's to come), and full video: ted.com/talks/keith_coleman_…

Great interview about how @CommunityNotes works! Super helpful for truth-seeking.

My interview from TED with @kcoleman and @audreyt just went live! Tons of gold in here: -why we stuck with crazy principles like not having an override button -how AI and humans collaborate to write better/more notes than either on their own -how much impact community notes are making -why our algorithm is smarter than regular upvoting/downvoting -new ideas for finding common ground on social media ted.com/talks/keith_coleman_…

Feel like the world is inescapably polarized? Think again! Talked frontiers of @CommunityNotes and finding Common Ground on social media with @audreyt and @jaybaxter at TED2026. There’s more agreement than it may seem, and it’s a reason for optimism! ted.com/talks/keith_coleman_…

Commentary is one of the most important pillars of X. And sometimes the best way to share your thoughts is with video. Today we're launching a whole new way to make them: React with Video Tap the repost button and start recording with green screen, split screen, or picture-in-picture. Now available on iOS

Incredible sign of progress that this is possible

Freeform explanations now available in note requests. Highly recommend typing something quickly here if you have a more specific thing you're wondering than "is this true?"

Many of you are vastly overconfident in Pangram

We can now harvest trees from the air energy efficiently. At current mission times, 6 drones match ground machine productivity. Autonomy lets one operator run multiple drones, scaling to make the world’s 1B hectares of production forest profitable, healthy and resilient.

Over the past month, we have identified a number of large accounts that have been programmatically reuploading content from smaller accounts to game the revenue share program and circumvent crediting the original author. We are now identifying these posts and allocating the impressions entirely to the creator. If you have insightful commentary about a post, we recommend using the Share Video or Quote feature to ensure your posts are properly attributed.

AI Note-writer Progress Our note-writer (wholesome-raspberry-stilt) has written community notes on X with 47M views. The cost per helpful note is about $7. You can look at our notes here: goodheart-labs.github.io/cn-…

Neat! Community Notes *do* seem to work. They reduce the number of reposts for posts that get them, cutting down on their reach and limiting the spread of misinformation!

🚨New paper in @NatureComms: Community-based fact-checking reduces the spread of misleading posts on X 🚨 👉Based on N=237,178 X cascades, we found that exposing users to CN reduces subsequent reposts by 61.2% 👉 CN increases the odds that users delete their posts by 94.3% 🧵

We heard you want notes faster Enter: note previews

📈📈

This FT article went way too viral... The study used no real humans. The simulations of humans are basic---LLMs prompted with political beliefs. They assume the synthetic human updates their political beliefs as a weighted average of their original position and the chatbot's response. I'd like to highlight some more exciting work by @MaxKronerDale @PReaulx @lukebeehewitt on "DeliberationBench" arxiv.org/abs/2603.10018 Chatbots will inevitably influence people. The question is whether that influence is procedurally legitimate. A useful comparison is deliberation: a process which we consider to produce procedurally legitimate opinion change, e.g., en.wikipedia.org/wiki/Delibe… In the "DeliberationBench" paper, they find that after talking to a chatbot, people on average, change their opinions similarly to how they would if they participated in a deliberation on the same topic. This provides some evidence that models may be producing epistemically desirable changes because they move people in the same direction that a process with procedural legitimacy does. Note that this doesn't directly answer whether the process the _model_ used was legitimate. A model could lie to the user but still get them to the deliberative outcome. To address that, we should try to directly align models with the deliberative ideals which give deliberation procedural legitimacy in the first place, something I argued for in my IASEAI'25 talk youtube.com/watch?v=NoM1BgZo…

AI community note writers are starting to drive serious scale we are starting to transition out of the minor leagues here

Gotta get the Bill Ackman phone out to read this note

All aggregators had their payouts reduced to 60% this cycle. We will add another 20% deduction in the next cycle. It became abundantly clear: flooding the timeline with 100 stolen reposts and clickbait everyday crowded-out real creators and hurt new author growth. The next step is to assign a permanent deduction to habitual bait posters who use “🚨BREAKING”on every post. X will never infringe on speech or reach—but we will not compensate for manipulation of the program or our users.