TED Talks Daily - How Community Notes reduce viral misinformation | Keith Coleman, Jay Baxter

Episode Date: June 10, 2026

Community Notes on X started with a wild idea: Instead of tech companies deciding what's true, what if you let people fact-check each other? Jay Baxter and Keith Coleman, who helped build the crowdsourced system adding context to misleading posts, discuss how the program reduces viral misinformation — and why people across the political spectrum trust it. In conversation with TED guest curator Audrey Tang, they discuss how their "surprising agreement" algorithm could reveal the common ground that quietly exists across a polarized internet. (Followed by a note from TED guest curators Divya Siddarth and Audrey Tang) Hosted on Acast. See acast.com/privacy for more information.

You're listening to TED Talks Daily, where we bring you new ideas to spark your curiosity every day. I'm your host, Elise Hugh. X, formerly called Twitter, is now using community notes, a crowdsourced fact-checking system. The company's algorithm architect Jay Baxter and its VP of product, Keith Coleman, built it. Starting with the question, what if the people got to decide what's true? If people don't trust tech companies to draw the line, could they draw it, then they'd You can download the real data, the community notes and ratings, run the code on the data to verify that there's no funny business that we're doing on our end.

Like there's no override button. So it's really by the people. It's an idea that has earned genuine interest and trust across the political spectrum, even as it's become entangled in a larger, more contentious debate about the dismantling of professional fact checkers. In this conversation, Jay and Keith sit down with Ted guest curator and civic technologist Audrey Tang to discuss how community notes actually works. If we can identify common ground at internet scale,

it'll make it a lot easier to create a future that humanity likes. They also talk about what they're working on next. And stick around. After the talk, we caught up with the TED guest curator's Audrey Tang and Divya Siddharth, who share a few more thoughts and takeaways on community notes for us to consider. That's all coming up right after a short break. And now our conversation of the day. You know, we built community notes because we wanted to build a better informed world.

And as it scales to more parts of the Internet, that means more people have access to accurate information. Great. So let's look at a community note. That's a note. So what is it? Yeah. So this is a real example of a community note we're looking at. So basically here, the post on the left is about Iran. And it's saying the USS Lincoln has been damaged and there's casualties. but actually the image is AI generated. So this thing on the right here that says readers added context, that's that people might want to know. That's a community note.

And what it's doing there is it's actually giving a lot of specific details about what's wrong in the image. And it turns out that that level of detail that it goes into is a big reason why people on both sides of the political spectrum actually trust community notes more than a generic misinfo warning. The way these get here in the first place is they're actually written by a regular user, a community notes contributor. And before they show on the platform to everyone,

on and attach on the post, they are rated helpful by people from different perspectives, so they're not shown unless that happens. Another quick thing to call out is, actually, a lot of the best notes are not just fact checks. They can add context to posts that are correct, but otherwise misleading. Okay, so it's a context engine for news, but is it also for official accounts or ads or any kind of post? Yeah, so a really important principle of the program is that all posts are eligible, That means post from Heads of State, posts from our company can get noted. As Elon likes to point out, his posts get noted.

It regularly identifies AI-generated imagery. It's been a ton of that recently with the Iran conflict. It's detected deep-fake audio of world leaders. It covers lighter subjects like entertainment, fashion, et cetera. We've even had multiple notes on both recent White House administrations. And at least in one case, actually took down the post, issued an updated statement, and you can imagine, like, there was a person,

a random person on the Internet wrote that note. You know, this isn't like a famous person. They went out there, saw the White House said something wrong, typed in this note, put it up there, and then suddenly, you know, leaders of the free world changed their public statement. That's like a superpower for people, so you can see why they're motivated to contribute.

Yeah, well. So a teenager I heard, caused that rejection, and it really is a superpower. But what is the mechanism? Can you take us back 10 years ago before this superpower gets invented and distributed? What caused the invention? Yeah, I mean, the origin for me goes back to 2016.

I was just a Twitter user then. I was following the 2016 election. There were three televised debates that year, but every day there was a debate on Twitter. So that's where I was following. That's where the world was following. I remember getting a lot of good information, but it was also hard to tell what was true.

And I was thinking, just sitting on the outside, thinking, like, how is the world going to solve this problem? Like, damn, like, how are we going to do this in a way that works and that people feel is fair in a midst polarization? So then, fast forward three years. I was working at Twitter at that point, and the industry had tried a lot of stuff by then. Facebook had built a huge fact-checking program.

Twitter was working with fact-checkers, and we also had internal teams that would try to review posts and decide whether they were or were not misleading. And there were a bunch of issues with it. It was just very clear these solutions were not solving the problem. There were issues with speed. So typical fact checks, just to put it in perspective,

we're often coming back in two to four days, which is like infinity and internet time. scale was an issue. Typically, people could review, like, I don't know, 10 order of 10 posts or topics a day. And even if you could solve those, trust was the fundamental issue. There were just a lot of people who did not want

or trust tech companies to be deciding what was or was not accurate. And so I was managing a team at this time, handed that off and just went to go prototype crazy new ideas, one of which became community notes. Okay, so the crazy idea just to play back is to think from the bottom up, asking people to trust random strangers on the Internet

and amid a very high PPM polarization per minute environment, as you just alluded to, why would people trust random strangers? Yeah, it's a really good question, and it's when we got all the time getting started, but the reality is people do trust community notes on both sides of the political spectrum, and I think there's a couple big reasons why,

One is the process behind it. So it's totally open, transparent, verifiable. And this is pretty wild in the world of social media. You can actually download the real algorithm code that runs in production. You can download the real data, the community notes and ratings, run the code on the data to verify that there's no funny business that we're doing on RN. There's no override button. So it's really by the people.

I think secondly, the notes are just really good. So they speak for themselves. They tend to be really accurate. And the main reason behind that is I think the principle behind the algorithm that doesn't ingest any sort of external authority, it actually decides what notes to show by looking at agreement from people who have disagreed in the past. And sometimes we call that surprising agreement or bridging.

And one thing that's really cool about this algorithm, if you compare it to something like a more naive upvote-downvote system, like a majority rules type of thing, something like that would just end up showing really biased notes. And for here, our algorithm-rengths, takes advantage of partisanship and polarization. So for any community note on a polarizing topic, basically there's always going to be someone out there

who's really predisposed to disagree with that note. So before they're going to rate it helpful, they're going to go fact-check it from every angle possible. They're going to really check the sources in a lot of detail. And as a result, the notes that are actually found helpful in this way tend to be really accurate, tend to use primary sources, and tend to be pretty neutral in their language. Okay, so people on both sides, after turning polarization into essentially fuel, right?

Geothermal energy. Uplift something, and both sides are happy. But the person getting noted may not be very happy. So say a head of state gets noted, and the head of state happened to have the phone number of your CEO and just calls Elon and say, take it down by tomorrow morning. What would he say?

Yeah, so those emails like that, calls or whatever, they do come in. Fortunately, the answer is really simple. We have no override button. So if you're not happy with a note, you need to take it up with the people. And this was kind of a crazy idea when we started. We went into a room full of trusted safety people, and we're like, hey, so the notes that show are going to be the ones that the people decide, and we can't take it down.

There's no veto. There's no veto. Wow. And, you know, they're like, what? But like, right? Are you serious? What if there's a bad note? But I think, you know, the point was if it's the tech company's opinion, why is anyone

going to trust it? It needs to be the people's opinion. And so we stuck to that principle. Everyone got behind it. And yeah, we have no way of changing the status of a note. Okay. Which is wonderful. Okay.

It is wonderful. Yeah. And so what happens to the post after it gets noted? You can just see this thing's going super viral at the start, all the way up until it gets noted. Basically, after that point, it totally flattens out, gets almost no more views. And the kind of crazy thing about this is it's actually not getting down ranked by our 4U algorithm, the post. This is actually just because of what we call organic user behavior, where basically people are realizing now that the post is incorrect because the notes on it, so they're just

liking it less and reposting it less. So I think this is really cool. And one thing that I also love is because our data is totally open. Actually, a lot of researchers from around the world have looked into this and found the same thing. So people from Stanford, MIT, UW, and Paris and Luxembourg have all actually found a very similar thing that reposts will drop by about 50 percent or 2X after notes applied. And this is really big in the scale of social media, like one or five percent when would be pretty big

and the scale of typical A-B tests. So one thing that I think is really heartening about this is that we know from this and some other studies that actually people are not just entrenching their beliefs. When a note is applied to a post, they'll actually agree with the core claims in the post less. And I think that's really cool. And I guess there's a little bit of a mixed blessing here, though,

because actually post authors will also be more likely to delete their posts after they get noted. So in that way, the best notes actually get seen very infrequently. So I'm torn about that, because just for me personally, you know, I think not everyone agrees on this, but for me personally, I'd rather see a post and a note than neither at all, just because that's probably not the only time in the world where you're ever going to see that particular wrong claim,

so maybe you'll see it off X somewhere in another post. And just for me, seeing a lot of notes has kind of increased the skepticism that I have when reading things. Okay, they serve as inoculation, essentially. Yeah, it's kind of a big deal if this happens organically. People often assume the world is very polarized, Certainly it feels very polarized. But people here are just, they're just making a choice.

Where they see a post, they see a correction, they're like, yeah, things wrong. I'm just not going to share it. And that's happening across the political spectrum. And we've seen that pattern again and again. When we first were designing the products, we did interviews with hundreds of people, left and right. And it was really obvious that most people just want to know

what's going on in the world. They know they're consuming incorrect stuff. They just want to sift through it. And this is just showing that in action. When given information, they're going to try to make a good decision. And so I think people often assume, like, man, it must be tough to work on in the space of misleading information. It must be, you know, I get sad all the time, whatever.

It's actually, like, I feel very optimistic working on it because we see there's quite a lot of agreement. People are actually quite reasonable. Wow. Okay. So the PPM is going lower. It seems like it's lower than it might feel. Amazing.

So let me now push. on a more cynical take. Anyone who spent five minutes on the internet is probably thinking now there's going to be a way to game this, maybe many ways to game this. And just one example, I co-wrote a paper called

malicious AIS1. It talks about one person farming, like 5,000 agents, the machine kind, and then they are some coded left, some coded right, they behave completely normally, they contribute even to committee notes.

And just when the controversial issue or election happens, then they manufacture a surprising agreement and just note something that is actually true. How do you deal with that? Yeah, manipulation is a real thing. I mean, people are always trying to game social media algorithms, and community notes is no exception. So I think one thing to call out is that surprising agreement mechanism

does provide a bit of a defense against a more naive attack than when you described. There's a lot of people with the same view all piling on, trying to get an incorrect note showing, that's not going to work. But for a more sophisticated attack, like the one you describe, we do have a lot of defenses in place, so just to name a few,

we do things like requiring a verified phone number from a trusted carrier just to increase the probability that we're dealing with real humans. We look for raiders who have rated things really similarly in the past, and actually, we might treat them as the same person just to limit the influence of really similar behavior.

Another thing is we can look at random samples of raiders, and if they're rating things very differently than self-selected, possibly malicious raiders, and that's a very important signal. And we have other things to you, like there's raider reputation to deal with low quality people. But I think another key thing to call out is, even with all these defenses, you know, community notes are incorrect sometimes. Now, because it is really rare, we actually get the self-correcting property, where the incorrect notes attract a lot of attention.

And they'll draw a lot of raters to go quickly rate them not helpful, and then they'll stop showing. And I think that self-correcting property is super important also just in a breaking news situation, right? Something that was true a few hours ago may not be anymore. So it's great that notes are not set in stone. Okay, so notes being wrong is noteworthy. And so people recursively improve. Indeed, I've seen it happening on X for quite a while, because I was a long-time contributor.

It just feels like magic. It's like Wikipedia or Groghipedia. When many people swarm into some controversy, it just gets really nice. But what about the other situation in a niche topic, just developing fast? There's just not enough attention to bootstrap the initial surprising agreement. So I've also seen like five hours, ten hours go by without any consensus at all. And so how are you going to tackle this? because as we pointed out, if it comes next day, it's already gone.

Yeah. Well, first, just to level set on speed, I think Keith already mentioned, you know, the previous study of the art fact-checking would often take on the order of days, and community notes is usually more in the order of hours. So it is already quite a bit faster. Notes can appear as often as about 20 minutes on a brand-new post, but they can actually appear instantly if there's already another note out there that's matching on a URL or image or video. I think on top of that, one thing that people really, really like is if someone actually sees a post and engages with it

before a community node is appearing, we'll actually send them a push notification later, so they get the correction as soon as the note comes out. Now, even with all that, I think it's super important for us to keep making community notes faster. People want instant context, and rightfully so. So to that end, what we've done last year is we've actually opened up an open API for AI contributors.

And this is a little bit of a crazy thing, and the totally open spirit of community notes, just like a regular person can be writing notes. We let regular people write their own AI note writers and submit notes to our system. And what we've seen so far is it's actually working really well. The notes are really fast, and they're quite good. But definitely, because it's AI, they're wrong some of the time.

So the way we treat this now that's been working well, is we still have that human layer, where humans rate the notes in the same way as any other human-author note. And what we're working towards now is a way for AI and humans to collaborate more effectively to co-write better notes faster. So humans are not just downloading or upvoting,

but working with AI models. Yeah, the idea is they can... Like, can we have humans and AI co-write these things together, co-create them together? And does that allow us to do this at a much faster speed and larger scale? If there's demand for a note, like people who are requesting a note on a post, AI will take a first shot at it.

Humans can write to you, but A.O. will take a shot. And a bunch of people, they rate it and they make these suggested improvements. They can also leave suggestions on style or tone, so they can say, like, hey, I think this source is biased, or I think you should use a primary source. It's going to be more trustworthy. And then the AI takes that, regenerates a note, and, you know, usually gets it right. And what's cool about this is, first you get to... a better note on this post people care about.

But two, all of those corrections, all the suggestions are training data that you can feed back into the AI. So you can make it less likely to make that mistake again. You can make it better at researching in the first place, and also you can make it more neutral, less biased. So all these human suggestions, both they make better notes and they make better AI. Okay, so just to play it back,

this is not GROC helping humans to such a degree that it takes over all the judgment costs. It is basically human teaching AI, collaborative learning, so that the translation between communities, like climate justice on one side, biblical creation care on the other, the AI model learns how to translate, and then what? They become better at this kind of translation?

Is this a new way to reward AI models? How does it work? Yeah, there's this thing that we sometimes call reinforce an warning from community feedback, as opposed to just reinforce learning from human feedback, which maybe would use potentially a smaller bias that of non-representative people. And basically in the case of community notes,

what it would look like is directly training the model to be writing notes that would be maximally likely to be found helpful by a simulated set of raters who typically disagreed in the past. OK. Well, that's really nice. Yeah, that's cool. Yeah.

So still, someday I just opened X. And I just see peak slop. Like the marginal cost for generating synthetic media, even synthetic intimacy, is now falling so fast. And so it's just sometimes I feel, personally, that whatever the corrective mechanism we invent is not going to be false enough for this kind of peak slop situation. So why should anyone here believe that computing that?

the company who knows and collaborative nodes will evolve to meet the demand? Yeah, so definitely in the last six weeks or so with the Iran conflict, we've seen the biggest surge in synthetic media that I've seen, at least in the kind of misleading info space. And I will say, like, we're on the frontier here. So this is the highest scale, highest speed solution that exists. These are new problems. So, you know, we don't know what's going to.

going to work, can't guarantee the problem won't be solved. But I think there's a bunch of reasons to be optimistic. For that problem, like synthetic media surge, we can both scale up the corrections and we can both change the fundamental incentives and dynamics of the system. So in terms of scaling corrections, we talked about AI, just to put some numbers on that.

In the last four months alone, we've doubled the number of notes that are showing on X. So that's like not trivial for a scale of service. It's a 2x and 4 months. I think there's clearly headroom on that. Is it 10x, 100x? I don't know. But there's clearly headroom to grow.

The other thing on the incentive side, one of the reasons people post these things is they can make money off of it through creator revenue sharing programs. And so we've recently put into place some changes to the policies there where if your post is noted, you can't make money off it. Also, if you post AI-generated, footage of a war or conflict, and you do not clearly call it out, you are suspended from the revenue-sharing program for three months, if you do it again, you're suspended forever.

And so that's like kind of a big deal. Those will shape the underlying motivations people have. That's huge. Okay. We've been talking about defense, defending against all those manipulations and engagement through enraignment and so on. But is there a future in which social media, instead of pitting human, against one another, puts people and connects them with each other,

elevating the voice that bridge. And you're like, we have just a demo. Yes, we are building this. This is an awesome future. So we have a pilot running. The idea is, so in community notes, we find the kind of like corrections or context that's helpful to people from different points of view. What if we could find the ideas or opinions that are liked by people from different points of view? And when it happens in the pilot program, the Post will just get a call out saying like by people from different perspectives.

And we see this. Obviously, people were very happy to see Delta not allow Congress to skip the TSA line until TSA was funded. And we see this, yes. Yeah, you're among millions of people who also feel this way. And we see this agreement across a lot of topics.

Things that you think of as controversial. We see it across immigration, across the economy, taxes, international conflicts, et cetera. There really is a lot of agreements out there. Not on everything, but there's quite a bit of it. And the concept is, like, if we can identify that, like, we don't need to boost this to start. Just, like, show people when there's agreement on something.

First of all, I think they'll find it interesting. It's a curiosity. Second, it might incentivize more of that. Like, maybe people will try to speak more in a way where they can find that agreement. And you get more momentum behind those ideas. Yeah, that's a really good point. I think just in the same way that community notes spread less,

even though there's no... Community Notes cause post-a spread less, even though there's no down-ranking in the algorithm. I think you'll probably see something in August here where there's just a positive second-order effect from making that common ground, common knowledge. So it's a common knowledge engine that turns polarization into what we can all live with.

This is truly visionary. And think about it, because this thing is open source. It is open data. So it means that not just X, but rather blue sky, true social, everybody, can just plug in that stream. And so that AI can learn from that and then connect the communities back together. So what if we apply this engine beyond social media?

Can you paint a picture of how that would look like? Yeah, so where my head always goes is imagine just for like one session of Congress, everyone just focused on delivering where there was agreement, you know, whether it's immigration taxes or whatever. I think people would be stoked. And yeah. Yes? I would be stuck.

There's a lot of agreement on these topics. If all we did was pursue the areas of agreement, I think people would be pretty happy with the direction the world was going. And so, you know, my hope is with programs like this, if we can. identify common ground at internet scale, it'll make it a lot easier to create a future that humanity likes. And so hopefully we can help with that. And with that, Jay, Keith, thank you for being our best builders and showing us that a pro-social media future is not in some sci-fi, it's already here. Thank you. Thank you. Thank you, Audrey. That was Jay Baxter and Keith Coleman

in conversation with Audrey Tang at TED 2026. And you, you know, You may have noticed we've been experimenting with something different on the show. We're calling it Curator's Corner. Throughout the year, you'll hear from Ted's curators. The people who actually find and work with the speakers you hear on the show, they will share more about the idea you just heard and the behind the scenes of how the talks come to life. And now here are TED guest curators Audrey Tang and Divya Siddharth.

I'm Audrey Tang, and I'm a guest curator along with the fantastic Divya Siddharth at TED 2026. As guest curators, we get to bring people who are doing incredible work into the TED stage, help them find ways to share that work with the world, and be able to create a dialogue between what we think are some of the best ideas out there and solving the problems we care about the most,

AI, democracy, these big questions, and the TED audience and really the wider world. And we chose the interview format to bring Keys and Jay in because we really feel that the 18-minute talk format as good as it is, it's not doing the full justice of their job, which is training in AI, to understand the differences between, say, climate justice communities and the biblical creation care communities and the various different aspects that this social translation can do to our democracy. So I try to push them like really hard in every answer they give and they took it like a champion.

I think one of the great things about this talk is, you know, a lot of it is about community notes, which is a fundamentally defensive approach, right? We understand that the world is full of lots of bad information. We try to prevent the bad stuff from spreading. But I love the ending, which is on, well, what would it look like if we flipped this? And I hadn't thought about that as much before, where if we flip this to say, as much as we know the kinds of corrections people agree on, we could also figure out the kinds of information and positive solutions people agree on and make that actually be the thing that people are focused on online instead of all the other stuff that they tend to focus on online.

This ending talks about data is soil, so that the understanding between different communities tend to get this garden of AI agents that grow with our communities, loyal to communities, and not trying to extract anything, but just to regenerate our deep understanding. If you're curious about TED's curation,

visit TED.com slash curation guidelines. And that's it for today. Daily is a podcast from TED. This episode was fact-checked by the TED research team and produced and edited by our team, Martha Estefanos, Oliver Friedman, Lucy Little, Emma Tobner, and Tonzica, Sungmarnevon. Additional support from Daniela Ballereseo, Christopher Faisie Bogan, Valentina Bohanini, Ban-Ban-Chang, Brian Green, and Laney Lott. Learn more atpodcasts.todd.com. I am Elise Hu. I'll be back tomorrow with a fresh idea for your feet. Thanks for listening.

Thank you.