Latest / Ctrl-Alt-Speech / A Midterm Night’s Dream

Transcript

A Midterm Night’s Dream · Ctrl-Alt-Speech

• 0:00
  Ben Whitelaw: So then, Mike, before we get into this week's big news in content moderation, internet regulation, tell me what fun thing you've seen on the internet this week
• 0:07
  Mike Masnick: So this one I am blaming on my children.
• 0:11
  Mike Masnick: So they have gotten me into this.
• 0:13
  Mike Masnick: Apparently, this is really big at the local high school.
• 0:16
  Mike Masnick: it is, a daily game called Krillion.
• 0:20
  Mike Masnick: Have you heard of this?
• 0:21
  Ben Whitelaw: No
• 0:21
  Mike Masnick: So I, I can't even quite describe it other than basically you have to answer seven questions.
• 0:27
  Mike Masnick: You have 25 seconds to answer each question, and it's just sort of semi-random questions, and the goal is to try and answer the most rare version that actually does fit the bill.
• 0:40
  Mike Masnick: So like the other day, there was something that was like, an object with, pages, right?
• 0:46
  Mike Masnick: And if you said a book, you would get the least amount of points because that everybody knows that.
• 0:51
  Mike Masnick: But if you came up with something much more clever and unique within the 25 seconds, you get more points.
• 0:57
  Mike Masnick: it is a little bit addicting in terms of like trying to think of the most unique thing to answer the seven different questions.
• 1:05
  Mike Masnick: It's, it's a lot of fun
• 1:06
  Ben Whitelaw: interesting.
• 1:07
  Ben Whitelaw: It's not dissimilar We have a show in the UK called Pointless, which is a, a bit like this.
• 1:12
  Ben Whitelaw: so it sounds like it would go down well in my household where we, we play Pointless quite a lot.
• 1:16
  Ben Whitelaw: and, uh, it's kind of fun for all the family, I guess.
• 1:18
  Ben Whitelaw: You, you can go up against your
• 1:20
  Mike Masnick: my high schooler is very good at this game, so I, I have not quite gotten there, as good as he is yet, so...
• 1:27
  Mike Masnick: What about you?
• 1:28
  Mike Masnick: What fun thing did you see?
• 1:29
  Ben Whitelaw: I've been using a really interesting app called Dial Flipper, which tries to solve the problem of podcast discoverability, which is very relevant to us, on the podcast.
• 1:39
  Ben Whitelaw: and it basically, it's a fun interface where you can essentially kind of, a bit like changing, the kind of dial on a radio, flip between different podcasts
• 1:49
  Mike Masnick: Oh
• 1:49
  Ben Whitelaw: that are similar to your preferences and, your kind of initial interest when you set up the app.
• 1:54
  Ben Whitelaw: So I put in obviously Ctrl-Alt-Speech, I put in a couple of other podcasts that, we refer to and, that I listen to regularly, and I've discovered kind of three or four others that are really fun and interesting.
• 2:05
  Ben Whitelaw: But it sends you kind of directly into an episode, so you're not necessarily going for right from the beginning.
• 2:11
  Ben Whitelaw: It drops you in.
• 2:12
  Ben Whitelaw: It's a bit like as if you were changing channel, on the radio essentially.
• 2:16
  Ben Whitelaw: So kind of using like an old analog, way of consuming information and applying it to, to new forms.
• 2:22
  Mike Masnick: Conceptually, I like it.
• 2:24
  Mike Masnick: Though if you are listening to us on Dial Flipper, don't flip the dial on us.
• 2:29
  Mike Masnick: But, that's a, that's a great idea for, for podcast discovery.
• 2:33
  Mike Masnick: I listen to too many podcasts already, so I'm a little afraid to try it'cause then I'll get hooked on others.
• 2:38
  Mike Masnick: But, uh, that's very cool.
• 2:39
  Ben Whitelaw: Yeah.
• 2:40
  Ben Whitelaw: And a good reminder as ever to, do the old-fashioned means of discovery listeners and rate and review us wherever you get your podcasts.
• 2:46
  Ben Whitelaw: it ma- makes a massive help, so do that, then do some dial flipping as well Hello, and welcome to Ctrl Alt-Speech, the podcast where we make sense of the major debates shaping online speech, platform power, content moderation, and the future of the internet.
• 3:07
  Ben Whitelaw: It's October the 29th, 2026, and this week we're talking about Lula going after big tech in Brazil, what we can expect in the US midterms next week, and for our extended Patreon episode, what Ofcom thinks about age assurance.
• 3:19
  Ben Whitelaw: Ctrl Alt-Speech is the very squeaky clean face of Mike Masnick, founder and editor of Techdirt, and me, Ben Whitelaw, founder and editor of Everything in Moderation.
• 3:28
  Ben Whitelaw: I don't know if, if we would ever get elected, Mike.
• 3:31
  Ben Whitelaw: Uh, we're gonna be talking a lot about elections this week.
• 3:34
  Ben Whitelaw: Ever thought about running
• 3:35
  Mike Masnick: No.
• 3:35
  Mike Masnick: No.
• 3:36
  Mike Masnick: I, I, I decided long ago that my...
• 3:38
  Mike Masnick: I mean, it, it is probably less true these days than it was in the past, but I decided long ago that my written history of things I have written sort of made me, deliberately unelectable so that I, I-- it would never even cross my mind to run for election because there were so many things that people would pick apart and attack me for.
• 3:57
  Mike Masnick: and yet these days it does feel like you can say all sorts of crazy things and still get e-elected.
• 4:03
  Mike Masnick: but still in my head, I have decided that I have made myself unelectable and I would never never consider
• 4:08
  Ben Whitelaw: tell us what the, tell us the thing that you would make you unelectable.
• 4:11
  Ben Whitelaw: I'm desperate to
• 4:12
  Mike Masnick: I-- just everything.
• 4:13
  Mike Masnick: I mean, my views on everything, in over 50,000 Tecterd posts I think just make me...
• 4:19
  Mike Masnick: You know, I, I just picture, opposition researchers reading through all of my articles on Tecterd and finding things that they can use against me and, you know, There's-- I'm sure there's a million things that could be taken out of context to attack me.
• 4:33
  Ben Whitelaw: For m- for me it would be the, you know, unending stream of photos that I posted on Facebook in my university years, all of which are still probably, available.
• 4:42
  Ben Whitelaw: Do not go and listen to those, listeners.
• 4:44
  Ben Whitelaw: Um, we've got loads to crack through this week, Mike.
• 4:46
  Ben Whitelaw: We've got, three I think really quite chunky stories to tackle.
• 4:50
  Mike Masnick: I did wanna say before we do that, next week for Patreon listeners, we have a special episode coming.
• 4:56
  Mike Masnick: Both Ben and I are unavailable, so we will not have a regular episode, but now is a very good time for you to subscribe to the Patreon version if you want our special bonus episode that we'll have.
• 5:07
  Mike Masnick: We will have a shorter trailer version of it on the main feed, but it is an episode mainly for our Patreon supporters
• 5:15
  Ben Whitelaw: Exactly.
• 5:16
  Ben Whitelaw: not only do we have that special episode for Patreon supporters, but today's extended episode is, ripe for anybody who's been following the global move towards social media bans or enjoyed our recent episodes about ID verification and some of the, clusterfucks around that, in all honesty.
• 5:33
  Ben Whitelaw: you will want to listen to today's full episode.
• 5:35
  Ben Whitelaw: It's, ripe for you.
• 5:36
  Ben Whitelaw: not only do you get today's episode in, in its extended form, but you get all future episodes, and you get those a day early compared to the free feed.
• 5:45
  Ben Whitelaw: So, lots of reasons to, to go to Ctrl Alt Speech Patreon page, patreon.com/ctrlaltspeech, and become a supporter.
• 5:54
  Ben Whitelaw: Mike, we're gonna start with one election.
• 5:56
  Ben Whitelaw: We're gonna then move on to, one that's happening next week.
• 5:59
  Ben Whitelaw: But we, we can't not mention, I think, what happened in Brazil over the weekend.
• 6:04
  Ben Whitelaw: It was a few weeks ago that Katie Harbath, who formerly worked at Meta on election integrity, has written a book all about it, came onto the podcast and said, Everything can change between the first and second vote in the Brazil election." little did I know how sh- how true she would be.
• 6:20
  Ben Whitelaw: President Lula claimed a famous victory, added, uh, something like 10 million votes on, from the first vote to the second vote, and subsequently beat, his Bolsonaro opponent, who's now, gone AWOL, I think, and um,
• 6:35
  Mike Masnick: to my fair country, yes
• 6:37
  Ben Whitelaw: gone to seek refuge with, Uncle Donald.
• 6:39
  Ben Whitelaw: but very interesting for us, Lula, in his, victory speech, promised to go after big tech platforms, which have been a feature in the election campaign, we've talked about it on the podcast, and, was pretty kind of vociferous in what he was planning to do.
• 6:54
  Ben Whitelaw: He said,"We cannot have monstrous companies earning billions, paying no taxes, and trying to influence elections." a very, very stark warning to tech platforms who are, based in Brazil, which which is obviously a big market for many of the platforms.
• 7:07
  Ben Whitelaw: what did you make of the kind of shift from vote one to vote two?
• 7:11
  Mike Masnick: Yeah.
• 7:11
  Mike Masnick: I mean, I'm not an expert in electoral politics in, any sense, and certainly not in Brazil, but it, certainly was a big deal.
• 7:19
  Mike Masnick: Apparently, it's the first time that there's been such a shift from vote one to vote two.
• 7:24
  Mike Masnick: and it does seem clear that, you know, the internet had an impact in some form or another, which is interesting.
• 7:30
  Mike Masnick: I mean, the, discussion that a lot of people had was that, there was, uh, I believe a Brazilian congressman who built up a, huge WhatsApp, following, WhatsApp group following or channel, promoting Lula and, trying to get out the vote and really trying to mobilize people in response to the sort of weak first vote.
• 7:49
  Mike Masnick: And, you know, there are debates over how impactful that was, but, it had, I think it was 10.8 million followers of that channel, at the time of the election.
• 8:00
  Mike Masnick: And, uh, Lula certainly credited the guy who ran it with, really, being a change maker and, and making a big difference in terms of getting out the vote for the second election.
• 8:10
  Mike Masnick: I did think it was interesting that, as Lula is praising the success of this one WhatsApp group, he is at the same time then saying we need to heavily regulate Meta, the company that owns WhatsApp, because there was this big controversy that went on basically over the last month about various Meta content moderation decisions that many people felt, you know, were unfairly targeting Lula and his supporters.
• 8:37
  Mike Masnick: there are all sorts of open questions on that.
• 8:39
  Mike Masnick: Meta put out a very interesting news release, I guess, in Portuguese, which, was targeting people in Brazil and basically said, that nothing was deliberate.
• 8:50
  Mike Masnick: They weren't taking sides in the, in the election there were a couple of different mistakes, and most of them, as people have spent a lot of time looking at how trust and safety and content moderation works, I would argue again, like I have in the past, that, most of the, explanations of what happened sound legitimate.
• 9:08
  Mike Masnick: They are, you know, small little mistakes or they are attempts to comply with existing regulations.
• 9:15
  Mike Masnick: I mean, there was one example of, you know, that you couldn't generate AI imagery of any of the candidates, which was following a Brazilian law that said as much.
• 9:27
  Mike Masnick: and there was a, a sort of narrow loophole that some people figured out where if you didn't put in the candidate's full name, but you put in their first name and their, election number, which I, I found out, I had a, discussion at a, one of the conferences I was at recently with a, guy from Brazil that, like, election numbers are a big deal, and they, they have these cards, like you can't bring phones into the election booth with you, so people hand out cards with the numbers of who it is you're supposed to vote for, which is kind of this, this interesting thing.
• 9:59
  Mike Masnick: He was showing me cards, of different, you know, candidate slates.
• 10:03
  Mike Masnick: and so but if you put in the number and first name, which people were doing with Flávio Bolsonaro, it would generate those images.
• 10:10
  Mike Masnick: you know, so people were saying,"Oh, they're biased.
• 10:12
  Mike Masnick: You know, they're allowing you to do AI-generated images of Bolsonaro but not Lula." a-again, most of these seem like the kinds of things that happen in the process of trust and safety and content moderation, they do their best and there are mistakes being made.
• 10:26
  Mike Masnick: But, it's tough when you look at how many of these.
• 10:29
  Mike Masnick: Like, I understand why everyone looks at this and says like,"Oh, it was Zuckerberg.
• 10:34
  Mike Masnick: You know, he's, buddied up with Donald Trump, and Trump wanted Bolsonaro to win, so now he's like putting his thumb on the scale." I find it difficult to believe that is what actually happened, but I understand why people look at that and react that way.
• 10:49
  Ben Whitelaw: Yeah.
• 10:50
  Ben Whitelaw: I I've got to say, like, I, I'm so fed up of this technical glitch excuse, right?
• 10:54
  Ben Whitelaw: This, this has, this has been something that have, you know, has been parroted phrase of platforms the last two decades, basically.
• 11:02
  Ben Whitelaw: and I, I wrote something for Everything Moderation I think in 2018 that's like, technical glitch is not a good enough excuse.
• 11:08
  Ben Whitelaw: And, and, and that, that obviously doesn't kind of underestimate how hard content moderation to do at scale is.
• 11:15
  Ben Whitelaw: But the, the response, the press statement or whatever the kind of public, reaction to is that from a platform is just feels so inadequate.
• 11:25
  Ben Whitelaw: And I think now that we-we're starting to see better transparency from actually some of the AI platforms around errors and, you know, kind of hallucinations and, and some of those kind of reports that are coming out are much more, they're much more transparent, they're much more detailed, they're much more to the kind of core of what the issue is, and they make re-reactions like we've seen from Meta and others in these situations look completely inadequate, frankly
• 11:50
  Mike Masnick: Yeah.
• 11:51
  Mike Masnick: that I totally agree with.
• 11:52
  Mike Masnick: I think, I think it is fascinating to see how willing the AI companies, Anthropic and OpenAI in particular, to do very, very transparent deep dives.
• 12:02
  Mike Masnick: It's, almost-- It's not quite the same, but it's, almost to the level of like when there are airplane accidents and the willingness of the companies to be totally transparent about here is exactly what happened, doing a deep dive, in an effort to make sure that the same thing doesn't happen again, where they're willing to open up and be entirely public about the mistakes that were made.
• 12:24
  Mike Masnick: yeah, I think that's fair.
• 12:25
  Mike Masnick: Like, it would be much better if social media companies were willing to go to that level of transparency as well.
• 12:31
  Mike Masnick: The other thing that was interesting that I saw was, you know, a group of Meta employees came out with a letter demanding that the company, explain itself for why it was, you know, the employees sort of indicated that they felt that it was tampering with an election or interfering with the election.
• 12:46
  Mike Masnick: you know, I don't think that they were employees who would be in a position to actually n-know that, but just the fact that employees were willing to speak up and say,"Hey, this looks really bad, and we want Meta leadership to explain themselves," I think is an interesting, moment as well.
• 13:01
  Ben Whitelaw: Yeah, we're seeing, we're seeing a bunch of organizing, probably on the back of this very large WhatsApp channel by Brazilians around Meta's work during this election.
• 13:11
  Ben Whitelaw: And not only have we got that letter that's come out, we've also seen kind of in-person protests, which, are kind of all wrapped up in a very good piece by Tech Policy Press that people should go and have a, have a read of.
• 13:21
  Ben Whitelaw: uh, 60 or 70 people in London, outside of the Meta headquarters in central London, 1,000 people outside of Meta's São Paulo office.
• 13:30
  Ben Whitelaw: you know, I think it's been a while since we've seen that kind of in-person turnout outside of offices.
• 13:36
  Ben Whitelaw: and I think those kinds of actions do have effects on employees.
• 13:41
  Ben Whitelaw: you know, if, if you're struggling to get into your office because you've got a bunch of people who are frustrated about the work you've done or what your company's done, you start to realize, you know, it becomes very concrete and very tangible, what your platform can do to people around you.
• 13:54
  Ben Whitelaw: Everyone talks about tech platform employees being a little bit detached from reality, um, and that I think can shift the thinking a little bit.
• 14:00
  Ben Whitelaw: So it's been really interesting to see that happen over the last week as well.
• 14:03
  Mike Masnick: Yeah.
• 14:04
  Mike Masnick: You know, and, Brazil has always had sort of an interesting relationship with, with tech regulation.
• 14:08
  Mike Masnick: I mean, I forget exactly when, but a decade ago or a little over a decade ago, you know, they passed this, Marco Civil law, which was, a sort of, uh, CDA 230 kind of thing for Brazil.
• 14:21
  Mike Masnick: it was very speech supportive, and I thought it was a very strong law.
• 14:24
  Mike Masnick: That has been chipped away continuously since then.
• 14:27
  Mike Masnick: But there have also been all of these situations where you get the feeling that officials in Brazil, and certainly judges in Brazil, feel somewhat powerless against the platform since they're not based in Brazil.
• 14:41
  Mike Masnick: And there have been cases, you know, going back a decade or so where judges have ordered...
• 14:47
  Mike Masnick: I remember WhatsApp, employees, like they were trying to regulate some aspect of WhatsApp and they wanted to jail, like a high-level Meta employee in Brazil basically because WhatsApp wouldn't shut down some, certain content.
• 15:01
  Mike Masnick: So there's always been this, this sort of weird, powerlessness in Brazil, especially among judges, where they feel like they wanna regulate what is allowed and what is not allowed on social media.
• 15:12
  Mike Masnick: And, it's been a little unclear what they could do, and so they would resort to things like ordering employees to be sent to jail.
• 15:19
  Mike Masnick: So it will be interesting to see what sort of actual regulations come out of this.
• 15:23
  Mike Masnick: I'm always a little bit nervous about, you know, when you have Lula getting up there and giving this victory speech saying,"We're gonna heavily regulate Meta," that often leads to bad regulations and ones where it is effectively whoever is in power now has the ability to censor content they don't like.
• 15:40
  Mike Masnick: and that would be an unfortunate result, I think, in, Brazil, whether or not you support Lula or Bolsonaro.
• 15:47
  Mike Masnick: but it is a challenge.
• 15:48
  Mike Masnick: We live in this world now where, you know, a lot of people seem to feel that, free speech is good if it is supporting the side that you support and it's bad if it's the other side.
• 15:58
  Mike Masnick: And that makes it very tricky for there to be any kind of intermediary whose job is hosting speech in some form or another, because they become this kind of lever of how do you control the kind of speech that you allow, and how do you promote the kind of speech that, that you want to support, and how do you suppress the speech that you claim is problematic?
• 16:17
  Mike Masnick: And in some cases it may be problematic, but in some cases it may just be, political speech on the other side of, the aisle, and that, that becomes very tricky to distinguish.
• 16:25
  Ben Whitelaw: Yeah, yeah.
• 16:27
  Ben Whitelaw: so that, that's one election.
• 16:28
  Ben Whitelaw: We have another set of elections happening in your fair country.
• 16:32
  Ben Whitelaw: And maybe, maybe Flavio was simply getting a flight to your fair nation to get a front seat at the midterms.
• 16:38
  Ben Whitelaw: No other reason than just to, kind of see how it all pans out.
• 16:41
  Ben Whitelaw: We can't speculate on that.
• 16:42
  Ben Whitelaw: But, obviously the ne- the midterms are next week in the US.
• 16:46
  Ben Whitelaw: have been a number of stories and reports about platform readiness for these elections and the impact of, of AI or potential impact of AI on, what's gonna happen.
• 16:56
  Ben Whitelaw: There's been a whole-- There's been a lot of people who've been voting early in the midterms, and there've been a number of stories about, mis- and disinformation around that as well.
• 17:04
  Ben Whitelaw: Generally, how are you, you know, you're, you're obviously there.
• 17:07
  Ben Whitelaw: How are you feeling about it?
• 17:09
  Ben Whitelaw: What is your sense from, from people you speak to?
• 17:11
  Ben Whitelaw: Can you give us a readout of that?
• 17:13
  Mike Masnick: Yeah, I mean, I think, there's concern.
• 17:17
  Mike Masnick: I don't think, You know, I, I mean, this happens every election, right?
• 17:20
  Mike Masnick: there's concern and, you know, the American elections, we have reasons for concern.
• 17:24
  Mike Masnick: We had someone who wanted to deny the results of an election in the not too distant past.
• 17:30
  Mike Masnick: and so there's that.
• 17:31
  Mike Masnick: But there is also this sort of immediate, you know, sort of knee-jerk reaction to everything.
• 17:36
  Mike Masnick: So, like, the big story yesterday as we're recording this was that if you were in the US, not if you were outside the US, but if you were in the US and you did a Google search on who won the 2020 election, the AI results, within the US in Google, in Chrome only, I think, would not tell you who won the 2020 election and instead told you to look up the results, at an official site.
• 18:03
  Mike Masnick: Whereas if you asked for any other, presidential election, it would tell you the results.
• 18:08
  Mike Masnick: If you asked Gemini directly, it would tell you the results.
• 18:10
  Mike Masnick: There were all sorts of things, but it, it refused to say who won the 2020 election.
• 18:15
  Ben Whitelaw: Techni-technical glitch?
• 18:16
  Mike Masnick: Uh, yeah, potentially a technical glitch.
• 18:18
  Mike Masnick: Google did come out last night and basically said,"Yes, technical glitch we're fixing." and again, like going back to what you were just saying, like, yeah, I do feel like that is an unacceptable answer, and it would be nicer if they explained why that kind of technical glitch happened.
• 18:35
  Mike Masnick: And I can, I can make guesses as to why that kind of technical glitch would happen, which is that there is a ton of information online, and much of it deceptive and, you know, purposely deceptive to try to convince people that Joe Biden did not win the 2020 election when he obviously did
• 18:52
  Ben Whitelaw: Yeah.
• 18:52
  Ben Whitelaw: 'Cause you, you wrote a bit about this this week for Tech There, right?
• 18:55
  Ben Whitelaw: And so, so just kind of unpack what you mean by, source of information kind of being deceptive and, and how you think that might affect this week,
• 19:03
  Mike Masnick: so this was based on a few different articles.
• 19:05
  Mike Masnick: I, I wrote a, a piece that sort of pulled together a bunch of different threads, about the sort of AI impact on, on the elections.
• 19:12
  Mike Masnick: And one of the things that was found in a piece by Will Oremus at"The Atlantic" specifically was that if you were asking chatbots about election information, there was a, high likelihood that it would give you some bit of false information.
• 19:27
  Mike Masnick: And there are all sorts of questions about like, how impactful is that false information?
• 19:31
  Mike Masnick: Is it important false information or, or what?
• 19:34
  Mike Masnick: But the AI chatbots would often return false information about candidates and about elections, and that is a concern.
• 19:41
  Mike Masnick: And the, point that I was trying to raise was that some of this may be due to the fact that the sort of mainstream, more trustworthy publications have, in many cases, decided to block themselves off from AI tools unless those tools are willing to, pay a large sum of money.
• 20:01
  Mike Masnick: So right now, as we're speaking,"The New York Times" and a bunch of other news publishers are suing OpenAI.
• 20:08
  Mike Masnick: There's actually a bunch of filings coming in today and tomorrow, in that lawsuit.
• 20:12
  Mike Masnick: and there's a- you know, all these attempts to get licensing.
• 20:15
  Mike Masnick: And there was a study, this is from NewsGuard from a couple years ago, that basically said a huge percentage of, top-tier mainstream publications, I think it was around 70%, blocked AI in some form or another, but that over 90% of, low-quality information sources did not block AI.
• 20:36
  Mike Masnick: And when you have that big of a mismatch between which sources the AI tools are reading, it's perhaps not that surprising that the end result is that the AI itself is misinformed and therefore is then misinforming the electorate that is asking it questions.
• 20:51
  Mike Masnick: all of which is concerning, but it's not clear what exactly you do about that
• 20:56
  Ben Whitelaw: Yeah.
• 20:56
  Ben Whitelaw: Oh, I've got some theories.
• 20:58
  Ben Whitelaw: So, so fi- so first of all, the 2020 election, any of the big models would've ingested quality journalism that provided a verified answer to that question of who won the 2020 election prior to the publishers changing the way their websites could be scraped.
• 21:14
  Ben Whitelaw: that question is not recent enough.
• 21:16
  Ben Whitelaw: It's not something that happened last week.
• 21:18
  Ben Whitelaw: It is something that happened long enough ago that whatever model got asked that question should get that right, first of all, which then opens this question of like, okay, so what did go on?
• 21:28
  Ben Whitelaw: Why has this happened?
• 21:30
  Ben Whitelaw: It raises questions, it, it allows people to pose nefarious theories and conspiracies.
• 21:35
  Ben Whitelaw: Like, I don't buy it for that example.
• 21:38
  Ben Whitelaw: I think for, for more recent examples, things have happened since publishers started to kind of close down ranks against the models, and I've worked with a lot of publishers who have done this in the past, and advised actually a lot of publishers in recent years about how to deal with this, because, they need to make revenue and, and the revenue that comes from LLMs is a, tuppence on what they're able to get from direct revenue from, advertisers and readers.
• 22:02
  Ben Whitelaw: So, there's that.
• 22:03
  Ben Whitelaw: There's also the kind of the answer for me, and some AI labs are doing this, is to pay publishers for access to quality journalism that contains those answers, right?
• 22:14
  Ben Whitelaw: This wasn't in your piece and, I can see why, but, some labs are not doing that.
• 22:19
  Ben Whitelaw: some labs are.
• 22:20
  Ben Whitelaw: know, there are big deals that have been struck by OpenAI and Anthropic with big media orgs, News Corp, the FT, The Guardian, Der Spiegel, and others.
• 22:31
  Ben Whitelaw: So, there will be robust information in that about who won the 2020 election and a whole bunch of other things.
• 22:37
  Ben Whitelaw: Like, just be sure to prioritize that and not whatever wonky, AI-generated marketing blog a lot of the LLMs seem to cite for, a lot of answers, would be my, my thought.
• 22:50
  Mike Masnick: Yes, but again, like easier said than done.
• 22:52
  Mike Masnick: A-and like, yes, this is why it would be really good if Google came out with a full, explanation of what actually happened, because it is entirely possible that people within Google were like,"Oh, you know, if we answer that question directly about the 2020 election, then Donald Trump will get mad at us, and we don't need to deal with that headache, so let's just kind of, suppress that answer in some form or another." That's entirely possible, but it would be nice to know that for a fact rather than assuming it.
• 23:19
  Mike Masnick: But, you know, it's one thing to then say like, yes, you know, they should play up the more trustworthy information and they should designate that, but, again, the scale of this information Like it feels like, yes, with one answer that feels like it should be obvious, but with all of the world's information, how do you, how do you actually judge that quality?
• 23:40
  Mike Masnick: And there may be some systems that do that, but you're going to make some mistakes as well.
• 23:45
  Mike Masnick: It, it, there's a scale issue here.
• 23:47
  Mike Masnick: The other thing that does concern me about that, and, I've written before that I actually think the AI companies should be paying journalists, and my argument is that they should be funding journalism to keep the journalism going for, you know, they need more content.
• 24:02
  Mike Masnick: And I think, rather than licensing old stories, they should just be funding journalism across the board so that there's more new content for them to pull into their AI systems.
• 24:12
  Mike Masnick: I think that is something that they should seriously think about.
• 24:15
  Mike Masnick: Now, there are obviously then concerns about like, will there be honest reporting on the AI companies if they're funding journalism, which is a legitimate question as well.
• 24:23
  Mike Masnick: These are all things that have big trade-offs.
• 24:25
  Mike Masnick: But, again, when the systems are being flooded with deliberate misinformation, and there are all these attempts to sort of poison LLMs, you know, the Russian hackers that are putting out all sorts of information.
• 24:38
  Mike Masnick: somebody had discovered a, few months ago, like an entire think tank that didn't really exist that was just all AI-generated content clearly designed to push AI tools and chatbots to answer questions about Israel in a particular way.
• 24:56
  Mike Masnick: You know, there are all of these efforts to poison them, and if you have this situation where even if the companies come to deals with the news publishers, and even if they're, ingesting the high-quality information, that there's so much value for some people to pollute those answers that they're, you know...
• 25:15
  Mike Masnick: I, I mean, we've seen this in search engine optimization for years, and now it's an even bigger question when it comes to AI that there's enough motivated users out there who are trying to poison these things.
• 25:27
  Mike Masnick: And at the same time, if you have the higher quality sources then freaking out and opting out then, you're seeding the field to the people who are really trying to poison the results
• 25:38
  Ben Whitelaw: Yeah.
• 25:38
  Ben Whitelaw: we've seen obviously kind of, yeah, SEO become a massive thing.
• 25:42
  Ben Whitelaw: You would've had to deal with, changes to search rankings in your time as Techdirt, producing content for Techdirt.
• 25:49
  Ben Whitelaw: And I think we're gonna see a similar thing happen where generative, engine optimization, GEO as they're calling it, is gonna become a massive industry, and there's gonna be kind of black hat versions and people who are selling links and, basically milking the system and trying to, poison models in the way that you've explained.
• 26:06
  Ben Whitelaw: I guess we...
• 26:07
  Ben Whitelaw: the companies are moving so fast that that industry hasn't caught up.
• 26:10
  Ben Whitelaw: It doesn't seem that kind of the models have, changed the way that they're ranking content or evaluating content, quickly enough.
• 26:17
  Ben Whitelaw: again, that was something that, particularly Google did fairly frequently in the early days in, in the kind of early 2010s.
• 26:24
  Ben Whitelaw: So yeah, it's, clearly a bit of an issue.
• 26:27
  Ben Whitelaw: and that we, we don't know, I guess, what the impacts of these models are gonna be on people's voting.
• 26:31
  Ben Whitelaw: You made the point that actually it could be overblown, you know, how, whether somebody sees something in a summary on search or in a chatbot and then, you know, makes a decision about who to vote for on the back of it.
• 26:43
  Ben Whitelaw: I wanna talk a bit about kind of the old school mis- and disinformation piece as well,'cause there's a couple of stories that we'll include in the show notes related to that too.
• 26:51
  Ben Whitelaw: one of them is this kind of wild story about, James Tallarico Which I, you know, I kind of hadn't come across until I was reading stories for this week's episode, and this idea that he essentially is a Democratic candidate for the Senate in Texas, basically had the flu, but, you know, had this idea spun up by various elites, including Donald Trump and, Musk and a few other of, of those kind of henchmen, that he had been checked into a, a mental health institution, and it forced said mental health institution to put out a statement.
• 27:26
  Ben Whitelaw: Can you imagine the poor, the poor press officer who had to do that?
• 27:29
  Ben Whitelaw: Saying, saying James Talarico has not checked in to this institution, and Talarico to say,"I've actually never been to the state." Yeah, and it's just like almost like beggars belief, stories like that.
• 27:43
  Ben Whitelaw: it has nothing to do with kind of new technology.
• 27:45
  Ben Whitelaw: It has nothing to do with AI.
• 27:47
  Ben Whitelaw: It is people, from the top, political elites, business elites, technology elites, spinning ideas and, putting information into, vacuums and, and using content creators and influencers to amplify that in such a way that, yeah, you, you might think that this potential candidate has m- a mental health issue.
• 28:08
  Ben Whitelaw: what do you make of the kind of like, where are we with misinfor- disinformation in regards to the midterms?
• 28:12
  Mike Masnick: I mean There's always been mis- and disinformation around politics, right?
• 28:16
  Mike Masnick: This is the nature of political campaigning.
• 28:18
  Mike Masnick: There's another reason not to go into politics, by the way, uh, because, you know, you, you will see the worst version of yourself, in terms of how people present you on the other side.
• 28:27
  Mike Masnick: but like, there is that question of, of how convincing is any of this and how different is the fact that now we have AI tools to play into this.
• 28:36
  Mike Masnick: Now, there is the, there were some people who suggested that, with the Talarico story, there was, you this influencer who had made a phone call, claimed to have made a phone call to that institution where they claimed to have told him that Talarico had checked out, which just, it feels unbelievable.
• 28:53
  Mike Masnick: And some people then claim that that phone call was AI-generated as well.
• 28:57
  Mike Masnick: So there's possible that there was some AI-generated fakery involved, but I hadn't seen much follow-up on that.
• 29:03
  Mike Masnick: But the reality is that, political campaigns always spend time either spreading misinformation or, hinting at misinformation, right?
• 29:12
  Mike Masnick: This is, the nature of, you know, where we get too hung up on truth and falsity directly as opposed to just like implications and context where we, you know, if you imply stuff enough, you imply somebody has mental health problems or, or things like that, like that can influence people's vote, whether or not it's based on true facts or not.
• 29:30
  Mike Masnick: so yeah, I mean, I do think that this is just kind of the next level of this.
• 29:35
  Mike Masnick: I mean, it's the same thing that with social media.
• 29:38
  Mike Masnick: You know, rumors were spread on social media before you had AI, and some of those rumors were untrue as well.
• 29:44
  Mike Masnick: And then you get into the debate about like should the companies be policing that, and should their content moderation be pulling down rumors that aren't confirmed?
• 29:52
  Mike Masnick: this is partly the nature of political campaigning.
• 29:55
  Mike Masnick: There is an element in all of this which is that in theory, and people can disagree with this, but in theory, we tend to believe that in free societies, in democratic free societies, that we expect some level of responsibility, and agency on the individuals doing the voting to consume information and make up their own minds.
• 30:17
  Mike Masnick: And that leads to certainly an incentive for people to lie, but it should also lead for incentives for people to correct those lies.
• 30:24
  Mike Masnick: And we are seeing some, you know, it's a different story in Ohio, but there's also the, the Vex Ramaswamy, ad that he did w- with his, opponent's very much estranged mother, who was accused of helping, her then husband, who, was very credibly accused of raping, Amy Acton, which is Ramaswamy's op- opponent, when she was a 12-year-old girl, and then helping him escape Ohio.
• 30:51
  Mike Masnick: And they got her in this ad, which is backfiring.
• 30:53
  Mike Masnick: it's the kind of political trickery that is often common.
• 30:58
  Mike Masnick: You know, you use any tool that you have in the, in the drawer to, make your opponent look worse.
• 31:04
  Mike Masnick: And so getting her own mother to go on video and say that she's a liar, but apparently it's backfired in a big way.
• 31:11
  Mike Masnick: Like, as the real story behind her relationship with her mother came out and the horrible situation, and the fact that he would stoop to a level so low as to, effectively, like, find her living in a car, clearly she has serious problems, and to get her on video
• 31:26
  Ben Whitelaw: Right
• 31:27
  Mike Masnick: telling the story has backfired.
• 31:28
  Mike Masnick: And you know, there was a report I was reading this morning of people who claim to be, undecided voters who were felt swayed by the fact that, Ramaswamy chose to use that ad when it's such a problematic ad as well.
• 31:42
  Mike Masnick: And so, this is the nature of the information environment that we live in in a free society.
• 31:47
  Mike Masnick: Sometimes people lie and sometimes people mislead, and we sort of have to rely on people responding to that and getting the truth out there as best as they can.
• 31:55
  Ben Whitelaw: Yeah.
• 31:56
  Ben Whitelaw: I mean, that does, when you have candidates like that, Mike, it makes me think that maybe you've got a chance.
• 32:02
  Mike Masnick: Never.
• 32:03
  Mike Masnick: Never
• 32:05
  Ben Whitelaw: okay.
• 32:05
  Ben Whitelaw: Well, you know, this time next week we'll know, we'll know the results.
• 32:08
  Ben Whitelaw: Um, I'm sure we'll touch on it in, a future episode of, of Ctrl-Alt-Speech'cause we won't be, here next week.
• 32:13
  Ben Whitelaw: We've got that special Patreon episode.
• 32:15
  Ben Whitelaw: we're gonna touch on a, another interesting story that kind of broke just after we recorded last week's podcast with Jeff Horowitz.
• 32:21
  Ben Whitelaw: our producer Lee flagged this, I think it's really interesting, about France.
• 32:25
  Ben Whitelaw: France has been beset by widespread protests, across the country.
• 32:30
  Ben Whitelaw: you might have seen the pictures in your, feeds and on your news apps.
• 32:33
  Ben Whitelaw: the French government had been moving to, suspend social media accounts who were promoting protest violence, and that the kind of government had been asking social media platforms to, come in and have a, have a meeting with the view of kind of trying to quell some of that violence or, prohibit some particular speech on the, on those platforms.
• 32:54
  Ben Whitelaw: obviously a lot of interesting things are happening in France right now related to their own social media ban.
• 32:59
  Ben Whitelaw: The protests are, very much focused and, and driven by the kind of youth movement there.
• 33:03
  Ben Whitelaw: what do you make of, what's going on in, in France, Mike?
• 33:06
  Mike Masnick: Yeah.
• 33:06
  Mike Masnick: Well, I think it's actually, it's worth noting that it's more extreme than just saying, calling in the social media companies, which, the government did, and saying, they want them to ban certain accounts or, certain posts, which, you could kind of understand.
• 33:20
  Mike Masnick: I still have a problem with it.
• 33:21
  Mike Masnick: I don't think the government should be in the business of telling companies what content they need to ban.
• 33:26
  Mike Masnick: you could kind of understand it where you're saying like,"Oh, this is incitement to violence or, or whatever." But actually what they were talking about was even more extreme, which was the idea of entirely taking down social media services in certain areas.
• 33:40
  Mike Masnick: So saying like,"Oh, because we're worried that, protesters and looting may happen in a certain area, we want to have actual DNS blocking of a service in a particular area." So if you're in an area where the protests are, getting violent, turning off social media entirely.
• 33:58
  Mike Masnick: and that should be hugely concerning because that can clearly be abused in all sorts of ways.
• 34:04
  Mike Masnick: And this is happening the same week that there's a story that we mentioned very briefly, a few weeks ago that Quad9, which is a DNS provider, that they were, because of a, a legal dispute with sort of a sports streaming company that wanted Quad9 to be forced to block at the DNS level, and they-- That lawsuit, Quad9 effectively lost.
• 34:27
  Mike Masnick: They were told that they had to block, access to certain websites at the DNS level, and Quad9 announced that they were leaving France entirely because of this.
• 34:37
  Mike Masnick: And this is just an example of, governments and companies and courts, will go down the stack and try and find that centralized control point where they can, ban certain content.
• 34:50
  Mike Masnick: And at the DNS level, it's very, very broad.
• 34:52
  Mike Masnick: Part of Quad9's complaint was specifically that if they were to comply with this law, they would effectively be banning certain websites worldwide because that's how DNS works, even though they were only legally required to in France.
• 35:07
  Mike Masnick: And sort of the end result of that is like if in other countries they're suddenly required to ban at the DNS level, could have websites banned around the globe just based on accusations in one particular country, and you could see certain countries where that could be really, really problematic.
• 35:22
  Mike Masnick: And so Quad9 decided to step out.
• 35:25
  Mike Masnick: So I think like putting that into context and then saying having the social media companies called in to the government and being told like,"We want you to be prepared to pull down content or pull down entire websites or take down your social media, in certain cities within an hour if we suddenly demand you to do so," strikes me, even if you're concerned about the protests, even if you're concerned about the violence and apparent looting that is happening in some of the cases, like that strikes me as really a huge escalation in terms of government control and censorship on online networks.
• 36:01
  Mike Masnick: And so I, I find this story to be really, really concerning as to what's happening in France right now.
• 36:05
  Ben Whitelaw: Yeah.
• 36:06
  Ben Whitelaw: I mean, it, it kind of rhymes a little bit with you know that story we covered, I think probably about 18 months now ago, maybe longer, about the, bishop in Australia who was stabbed, and the video was on X and, the E Safety Commission tried to get it taken down globally to avoid kind of further incitement of violence.
• 36:28
  Ben Whitelaw: And Musk you know, I think decided to go to court with the E Safety Commission, and they ended up just dropping that case, not even that long ago, maybe, a few months ago.
• 36:37
  Ben Whitelaw: yeah, there's gonna be examples of, governments trying to control what's seen outside of their borders, I think increasingly.
• 36:45
  Ben Whitelaw: Why do you think that people aren't so concerned about DNS?
• 36:49
  Ben Whitelaw: Or, or why is it that there isn't legal cases being brought against, the French government in this case?
• 36:55
  Ben Whitelaw: Cause you could, you could see that there, you know, there should be a bit of a pushback against
• 36:58
  Mike Masnick: Yeah.
• 36:58
  Mike Masnick: I mean, there might still be.
• 37:00
  Mike Masnick: we'll sort of see what actually comes out of this.
• 37:02
  Mike Masnick: you know, it's interesting because this is all, some of the original fight, whatever, 14 years ago, 15 years ago about SOPA, PIPA, and the copyright fight was about this, right?
• 37:14
  Mike Masnick: The whole point of, the law in the US, SOPA, was that it was DNS blocking for copyright infringement, and a bunch of people, including infrastructure providers, pointed out that the dangers of doing DNS blocking, harm security because we have, DNSSEC, which is a, a security standard that a lot of, DNS services use, would effectively break, under these laws.
• 37:36
  Mike Masnick: and we're now seeing a return to pushing for DNS blocking, for copyright issues in the US as well.
• 37:43
  Mike Masnick: what it comes down to is that, people say,"Well, if we're just doing site blocking or having to sue the individuals, that's not as effective," and so they go a level deeper.
• 37:53
  Mike Masnick: And to me, the, the deeper you go into the stack, the more it leads to, government control of speech, and it should be really concerning.
• 38:01
  Mike Masnick: I do hope that there is legal pushback on this.
• 38:04
  Mike Masnick: You know, they choose things like this where they can point to, you know, violence in the streets and say,"We have to do something." But I, I worry about the larger impact of it and just the fact that, like, you know, we're seeing it happen in the copyright realm, and then suddenly it's like,"Oh, we can use that for this other thing where there's protests in the street." it strikes me as, really problematic.
• 38:24
  Mike Masnick: And, and, you know, if you compare it to, again, in that same timeframe, 2010, 2011 timeframe, where you had the Arab Spring, and you had people rising up and protesting and using Facebook for these things, if there were DNS blocking available to governments in the Middle East as the Arab Spring was going on, they would absolutely use it to shut down Facebook and shut down organizing for these protests.
• 38:47
  Mike Masnick: And so it's easy to say like,"Oh, we should be able to have government shut these, kinds of protests down," but, do you recognize what other kinds of protests may also be shut down in the process?
• 38:57
  Mike Masnick: And I think it's important to think through those implications.
• 39:00
  Ben Whitelaw: Yeah, there's probably some analysis to be done about Western nations liking it when it's done in, in parts of the world where they want to see change, but not, not liking it on the- in their own backyard.
• 39:10
  Ben Whitelaw: But that's for another episode, I'd say.
• 39:12
  Ben Whitelaw: Um, we're gonna wrap up today's free episode there, Mike.
• 39:16
  Ben Whitelaw: as I said at the top of today's episode, we're gonna go deeper on Ofcom's new rapid assessment of age assurance, technologies, which came out today.
• 39:24
  Ben Whitelaw: it's very, very fresh.
• 39:26
  Ben Whitelaw: We've been reading it, just before we started recording today's episode.
• 39:29
  Ben Whitelaw: We're also gonna wrap in, some other reports that the UK regulator put out last week, which we think have an interesting bearing on the kind of future of not only the social media ban in the UK, but maybe how other countries, follow suit.
• 39:42
  Ben Whitelaw: So tune into that if you're a Patreon supporter.
• 39:45
  Ben Whitelaw: If you're not, now is the time to become one.
• 39:48
  Ben Whitelaw: Go to patreon.com/ctrlaltspeech and we'll get started now.
• 39:54
  Announcer: That's the end of this week's free edition of Ctrl Alt Speech.
• 39:58
  Announcer: To hear the extended episode with additional coverage, become a Patreon supporter at patreon.com/ctrlaltspeech.
• 40:05
  Announcer: That's C-T-R-L Alt Speech.
• 40:08
  Announcer: If your organization is interested in sponsoring the podcast, contact us via our website at ctrlaltspeech.com.
• 40:15
  Announcer: Thanks for listening