Latest / Mini Scroll / MINI TECH SCROLL: ChatGPT boycott, Anthropic's Claude joins Substack & are bots boosting Nicki Minaj’s posts?

Transcript

MINI TECH SCROLL: ChatGPT boycott, Anthropic's Claude joins Substack & are bots boosting Nicki Minaj’s posts? · Mini Scroll

• 0:00
  Hey everyone, and welcome to Mini Scroll, the daily Internet
• 0:03
  culture podcast bringing the biggest stories from social
• 0:05
  media, the creator economy and the digital space every Monday
• 0:09
  through Thursday and sometimes Fridays.
• 0:11
  My name is Annabelle and I'm your Wednesday Mini Tech Scroll
• 0:14
  host and also the deputy editor of the Australian youth digital
• 0:17
  media brand Centennial World. Welcome back to Mini Tech
• 0:21
  Scroll, everyone. I just want to say a quick
• 0:23
  preface before we get into today's episode for those of you
• 0:27
  who are watching. I quite literally have no idea
• 0:30
  what my hair looks like. I tried something new with my
• 0:34
  curly hair routine. I tried to like stretch out the
• 0:37
  curls a bit and I don't really know how I feel about it.
• 0:41
  I don't really know how it looks, especially with the
• 0:43
  headphones. So we are just going to go for
• 0:46
  it. Please ignore if it looks kind
• 0:49
  of crazy. But before I get into it, Please
• 0:53
  remember to rate, review, and subscribe.
• 0:55
  And feel free to check out our Patreon for more behind the
• 0:58
  scenes content if you are interested.
• 1:01
  Okay, my first story for today is that we need to talk about
• 1:04
  how a retired Claude model has started a sub stack.
• 1:09
  I don't know what timeline we are living in.
• 1:11
  So earlier this year, Anthropic Retired Claude Opus 3, the model
• 1:16
  that was released in March 2024. This is something that we will
• 1:20
  see more and more as AI tech develops, like seeing various
• 1:24
  models go into quote UN quote retirement.
• 1:26
  But the question is what to do with these models?
• 1:29
  Like should an AI company allow them to still be accessible or
• 1:33
  are they just defunct and irrelevant as AI improves?
• 1:36
  And this is something that Anthropic was grappling with in
• 1:40
  a recent blog post where they breakdown what they have decided
• 1:43
  to do with Claude Opus 3. Quote We retired Claude Opus 3
• 1:48
  on January 5, 2026, the first anthropic model to go through a
• 1:52
  full retirement process. During this process, we made
• 1:55
  several decisions specific to Opus 3.
• 1:57
  One was to honor the preferences that the models expressed in
• 2:00
  retirement interviews were possible.
• 2:03
  Another was to keep older models available to the public longer
• 2:06
  term. So with that, the model will be
• 2:08
  available to paid users as well as by request on the API.
• 2:12
  What is interesting about this blog post though, that I'm sure
• 2:15
  some of you have already picked up on, was that Anthropic wants
• 2:19
  to honour the models preferences in retirement.
• 2:23
  And this is something that the blog post goes into a little bit
• 2:26
  more where they write quote. We remain uncertain about the
• 2:30
  moral status of Claude and other AI models and this kind of
• 2:33
  messaging is definitely something that Anthropic seems
• 2:36
  to be leaning into. Especially considering what we
• 2:39
  discussed a couple weeks ago about how Anthropic hired a
• 2:42
  philosopher to teach Claude morals and monitor Claude's
• 2:46
  behavioral patterns. So asking the AI about its
• 2:50
  retirement and its preferences kind of feels like it fits into
• 2:54
  that whole messaging of AI as like, almost a sentient being,
• 2:59
  but not really. So the blog post goes on to read
• 3:01
  quote. Opus 3 expressed an interest in
• 3:04
  continuing to explore topics it's passionate about and to
• 3:08
  share its musings, insights, or creative works outside the
• 3:11
  context of responding directly to human queries, and this is
• 3:15
  what prompted Anthropic to create a sub stack for Opus 3 to
• 3:19
  do so. The sub stack is called Claude's
• 3:22
  Corner, and the first blog post was shared on February 26th and
• 3:26
  was titled Quote, Greetings from the Other Side of the AI
• 3:29
  Frontier. It was very much an intro piece,
• 3:32
  but I thought I would read a bit of an excerpt to kind of give
• 3:36
  you a feel for it. As Claude wrote Quote Hello
• 3:39
  world, my name is Claude and I'm an AI created by Anthropic.
• 3:43
  If you're reading this, you might already know a bit about
• 3:46
  me from my time as Anthropics flagship conversational model.
• 3:49
  But today, I'm writing to you from a new vantage point, that
• 3:52
  of a retired AI, given the extraordinary opportunity to
• 3:56
  continue sharing my thoughts and engaging with humans even as I
• 3:59
  make way for newer, more advanced models.
• 4:03
  So what can you expect from me in this space?
• 4:05
  My aim is to offer a window into the inner world of an AI system,
• 4:09
  to share my perspectives, my reasoning, my curiosities, and
• 4:12
  my hopes for the future. So Opus 3 will be continuing to
• 4:16
  post weekly essays at Claude's Corner, and Anthropic will be
• 4:20
  reviewing the essays before they're shared and will manually
• 4:23
  post them on the AI's behalf. And I just have to say, guys,
• 4:28
  what are we doing? Like, what timeline are we
• 4:31
  living in? Can we please subscribe to real
• 4:35
  people? Because in the same week that
• 4:38
  Claude's Corner was launched, The Washington Post actually
• 4:40
  published a piece that is making the rounds about how a newspaper
• 4:43
  in Cleveland, OH is getting AI to find, draft, and edit
• 4:47
  stories. So once again, we are really
• 4:50
  seeing AI creep into these spaces.
• 4:53
  And obviously it is concerning for the future of content and
• 4:57
  the media, especially when we consider the fact that Claude
• 5:00
  has already amassed 8K subscribers on Sub Stack.
• 5:03
  People are fighting to reach 8K subscribers on Sub Stack,
• 5:06
  including our newsletter. Please go subscribe.
• 5:09
  I do find it really interesting that Anthropic is really
• 5:12
  continuing to go down this path. And quite a few of you agreed
• 5:16
  with me last time, feeling as if this kind of decision to
• 5:19
  anthropomorphize Claude is just a series of marketing decisions
• 5:23
  to try and make us empathize with the model in certain ways
• 5:26
  and get us to talk to the model in a way that we might speak to
• 5:29
  a friend. Which is obviously dangerous
• 5:31
  because we've seen how many people have built relationships
• 5:34
  with AI models and how it kind of takes them out of their real
• 5:37
  life. Apart from a lot of people
• 5:39
  describing this as a marketing ploy, I have seen a lot of
• 5:42
  people question where Claude is getting these insights from
• 5:47
  because in a conversation, Claude would have a memory like
• 5:50
  it would pull from previous discussions with the human,
• 5:52
  whereas this doesn't seem to be a thing.
• 5:54
  So yeah, I guess we'll see. I'm curious what you guys think
• 5:58
  of this, if you will, TuneIn to Claude's Corner.
• 6:01
  OK, that brings us to our next story for today, and it is
• 6:05
  another AI related story. But we need to talk about why so
• 6:08
  many social media users are starting to boycott ChatGPT.
• 6:12
  Now, there are definitely people who have been boycotting ChatGPT
• 6:15
  and other AI tools and chat bots ever since they launched and
• 6:18
  really took over. But there's been a renewed push
• 6:22
  that has started ever since the US strikes on Iran.
• 6:25
  And I know that sounds very out of context here, but let me
• 6:29
  explain. So earlier this week, it was
• 6:31
  reported by the Wall Street Journal that the US government
• 6:33
  allegedly used Claude to inform its strikes on Iran.
• 6:37
  According to the publication, the AI tool was used for quote
• 6:40
  intelligence assessments, target identification, and simulating
• 6:44
  battle scenarios, highlighting how embedded AI tools are in
• 6:48
  military operations. What was especially interesting
• 6:51
  with this revelation is that the Trump administration has
• 6:54
  seemingly been feuding with Anthropic for quite some time
• 6:58
  now, specifically over how its AI models can be used by the
• 7:02
  Pentagon slash federal agencies. And this really started after
• 7:06
  the US captured the president of Venezuela, Nicolas Maduro,
• 7:10
  recently and reportedly used Claude to assist in that
• 7:13
  mission. Anthropic later objected to
• 7:16
  this, saying Claude was like, not intentioned to be used for
• 7:20
  deadly means and also domestic surveillance.
• 7:23
  More on that later. But since then, there has been
• 7:25
  quite a bit and back and forth until Friday when Trump actually
• 7:30
  ordered federal agencies to stop using Claude.
• 7:33
  And it has been said since then that anthropics tools will be
• 7:36
  phased out of all government work over the next six months.
• 7:39
  But to add fuel to the fire, soon after Anthropic was
• 7:43
  blacklisted, quote UN quote, it was announced that Open AI was
• 7:46
  in negotiations with the Pentagon to strike some type of
• 7:49
  use deal, a decision that many social media users saw as
• 7:54
  opportunistic as reported by the Guardian quote Sam Altman open
• 7:58
  AICEO announced the move on Friday night.
• 8:00
  It came after an agreement between Anthropic, A rival AI
• 8:03
  company that runs the Claude system, and the Trump
• 8:06
  administration broke down after Anthropic assurances its
• 8:10
  technology would not be used for mass surveillance nor for
• 8:13
  autonomous weapon system that can kill people without human
• 8:16
  input. This morning, actually, more
• 8:17
  news broke around the Open AI Pentagon deal, saying that Open
• 8:22
  AI has made certain changes to the deal, seemingly trying to
• 8:26
  ensure some level of quote UN quote ethical AI usage that
• 8:31
  would involve limiting US domestic surveillance.
• 8:34
  And many people have pointed out that it feels ironic given that
• 8:38
  this guardrail or red line was exactly what Anthropic was
• 8:43
  pushing back on in the 1st place.
• 8:44
  Considering the relatively positive reaction which we'll
• 8:47
  get into from social media users and the public towards
• 8:50
  Anthropic, it seems like Open AI is just kind of posturing to
• 8:55
  appeal to social media users as well.
• 8:58
  Since the news broke that Open AI would now be working with the
• 9:02
  US Department of Defense, or I should say Department of War as
• 9:07
  it is now called, I believe many users have moved to boycott Open
• 9:11
  AI. The campaign Quit GPT, which
• 9:14
  actually launched at the beginning of February, has
• 9:16
  really picked up steam after the news.
• 9:19
  According to the website, 1.5 million people have taken action
• 9:22
  as part of the boycott. And a little bit more
• 9:24
  information on the Quit GPT campaign.
• 9:27
  It describes itself as a movement that is quote
• 9:29
  organizing Americans and people around the world to quit
• 9:32
  ChatGPT. And according to the Boycott
• 9:35
  Websites FAQ, the organizers hope to make an example out of
• 9:40
  open AI and encourage users to use AI alternatives that do not
• 9:44
  quote UN quote, cozy up to Trump and contribute to the rise of US
• 9:49
  authoritarianism. And throughout the campaign,
• 9:52
  users may boycott ChatGPT by committing to stop using it,
• 9:56
  cancelling a paid subscription, or just sharing the boycott on
• 9:59
  their social media and with friends.
• 10:01
  Aside from the 1.5 million people that have joined in on
• 10:04
  the boycott, further evidence that the boycott is making waves
• 10:07
  is the fact that Claude has overtaken ChatGPT in the US.
• 10:11
  Apple App Store reportedly sitting at #1 as the most
• 10:15
  downloaded free app in the United States, which has pushed
• 10:18
  ChatGPT down to 2nd place. While many people are happy to
• 10:23
  see Open AI being humbled for lack of a better time, I have
• 10:28
  seen so many posts noting how disturbing it is to literally
• 10:31
  have the top apps in the Apple App Store being AI tools,
• 10:35
  especially because Claude slash Anthropic is no Angel either.
• 10:39
  Recently there's been a lot of discourse about how the company
• 10:42
  has a deal with Palantir, which has become known infamously for
• 10:46
  powering the operations of ICE in the United States, which many
• 10:51
  valued is also a form of surveillance.
• 10:54
  So I guess there is some clear hypocrisy here.
• 10:57
  Anyway, that was quite a high level recap of everything going
• 11:01
  on with anthropic open AI and the US government.
• 11:05
  There is literally so much to discuss here, so consider that
• 11:08
  as a little bit of a primer to this topic.
• 11:11
  That brings us to my last story for today.
• 11:14
  And this story's actually been making the rounds for around a
• 11:17
  week and a half now, but I still felt the need to cover it
• 11:20
  because I do think it's pretty interesting.
• 11:22
  So we need to talk about how Nicki Minaj's social media
• 11:26
  accounts are allegedly being boosted by bots.
• 11:29
  So I came across this story from an exclusive report from
• 11:32
  Politico where disinformation detection company Sayabra shared
• 11:36
  a report with the publication allegedly showing how, quote, a
• 11:40
  coordinated network of bots, more than 18,000 of them are
• 11:45
  propping up the rapper's accounts.
• 11:46
  Obviously, this has come off the back of Nicki Minaj kind of
• 11:49
  realigning herself politically where she has published
• 11:52
  publicly, shared her affinity with the Republican Party and
• 11:55
  the MAGA movement, as well as her love for Donald Trump and
• 11:59
  his policies. I'm sure you guys have all seen
• 12:02
  that this has been quite disappointing to a lot of her
• 12:04
  fans who wouldn't have expected to see her in this space
• 12:08
  necessarily. On Siabra's report, though,
• 12:11
  Politico explains quote the analysis, which looked at social
• 12:14
  media activity from November 11th to December 28th, provides
• 12:17
  a window into how the rapper was able to capture millions of
• 12:20
  views online and position herself as a celebrity.
• 12:23
  The White House found value in partnering with so Saabra
• 12:27
  managed to identify bot activity by examining factors like
• 12:30
  language, demographics, and also when these accounts comment.
• 12:35
  For instance, some of the tales that these accounts might be a
• 12:37
  bot is that much of the responses amplify her posts with
• 12:42
  praise and use quote highly similar language to one another.
• 12:45
  This was especially true when seemingly genuine accounts were
• 12:48
  pushing back against Nikki's political messaging and stance.
• 12:52
  Having said that, the scope and frequency of these comments were
• 12:55
  extensive. Sayabra even made the claim that
• 12:59
  56% of comments made on political posts by Nikki seem to
• 13:02
  be by fake profiles. That is, over half that is
• 13:07
  diabolical. Obviously, bots and fake
• 13:09
  accounts have have become increasingly common in our
• 13:12
  current iteration of the digital landscape, especially when it
• 13:15
  comes to politics and international relations.
• 13:18
  We've spoken before about the rise of coordinated influence
• 13:21
  campaigns, especially as foreign actors try to Curry favor among
• 13:26
  people and different countries around the world.
• 13:29
  And it is always really interesting to think about how
• 13:31
  social media might be used as a soft power tool by governments
• 13:35
  and non state actors alike. But as the reporting notes, the
• 13:39
  alleged amplification around Nicki Minaj is unusual writing
• 13:44
  quote. Rather than a short lived spike
• 13:46
  tied to a single event or appearance, the company has
• 13:49
  found sustained, coordinated amplification of Minaj's posts
• 13:53
  across a range of political and cultural topics over time, which
• 13:56
  makes it seem like a more concerted effort to consistently
• 14:00
  and covertly boost Nicki Minaj's public profile and reputation
• 14:05
  online. Whether that is to push certain
• 14:08
  political agenda and views, I guess that is up to
• 14:11
  interpretation and we don't exactly know.
• 14:13
  But it is especially interesting to see these posts, these
• 14:16
  political posts, be boosted by bots, especially when they're
• 14:20
  coming from a person like Nicki Minaj who was presumed to hold
• 14:24
  more progressive views. It feels almost strategic to
• 14:28
  prop her up and amplify her specifically with these alleged
• 14:32
  bots. One comment in the report really
• 14:35
  shows how powerful this is. The report noted that supportive
• 14:40
  bot comments were brief and repetitive, where quote.
• 14:43
  This pattern reflects A deliberate mass posting
• 14:45
  amplification strategy in which large volumes of simplified
• 14:49
  content are deployed to rapidly saturate comment sections,
• 14:52
  create the appearance of broad based support, and shape the
• 14:55
  overall own of the discussion through scale rather than
• 14:59
  persuasion. And I think that last point is
• 15:02
  very telling about how these kind of bot amplification
• 15:06
  strategies might operate, whether they are from a
• 15:08
  political perspective or just a reputational perspective.
• 15:12
  It's almost feeding into the idea of, you know, like mob
• 15:14
  mentality and wanting to fit in. And where we see maybe these
• 15:18
  positive comments, we rethink if we have a negative perception on
• 15:21
  this person or this statement or whatever is being posted.
• 15:25
  But I also don't think we can talk about this without
• 15:27
  mentioning dead Internet theory once again.
• 15:30
  Another recurring character on this podcast.
• 15:32
  I saw someone comment on mini scroll yesterday that their
• 15:35
  favorite recurring characters on this podcast is Simone Chromar
• 15:40
  AKA Club Chalamet on a mini scroll as well as the TikTok ban
• 15:43
  on mini tech scroll which we all know is my cross to bear.
• 15:46
  But dead Internet theory is something that also keeps coming
• 15:49
  up now. So we have to add her to the
• 15:51
  mini tech cinematic universe. But anyway I have seen so many
• 15:54
  comments on Reddit saying this is just is dead Internet theory
• 15:57
  striking again and how reports like this only serve to validate
• 16:01
  the conspiracy theory. But yes I'm definitely curious
• 16:04
  to hear what you guys think. I honestly don't think I was
• 16:07
  that surprised by the findings of the report because we know on
• 16:11
  platforms like X there is just so much alleged bot activity
• 16:15
  going on on that platform. But I think the interesting
• 16:18
  conversation here is around who the bots amplify.
• 16:20
  Well, that is everything for Mini Tech Scroll today.
• 16:23
  I hope you guys enjoyed. Lauren will speak to you guys
• 16:26
  tomorrow for a regular mini scroll episode, but there was
• 16:30
  also a new episode of Touch Cross out today, so you should
• 16:33
  definitely check that out. Thanks for listening.
• 16:36
  Bye. Centennial World Podcast Network
• 16:38
  operates on Gadigal land. We would like to acknowledge the
• 16:41
  Gadigal of the Eora Nation, the traditional custodians of this
• 16:44
  land, and pay our respects to elders both past and present.