Latest / Connecting the Dots / AI Guardrails Tighten, Data Gets Valued, and Accountability Grows

Transcript

AI Guardrails Tighten, Data Gets Valued, and Accountability Grows · Connecting the Dots

• 0:00
  Today's tech news moves fast, but understanding
• 0:03
  it shouldn't be hard. That's why Snarfl Solutions
• 0:06
  Group brings you Connecting the Dots, where we
• 0:09
  break down what's happening in tech and why it
• 0:11
  matters to you. Now, let's dive in with your
• 0:14
  hosts. And welcome to Connecting the Dots. I'm
• 0:17
  Alex. And I'm Morgan. It is Friday, January 16th,
• 0:21
  2026, and we are looking at a really fascinating
• 0:25
  split happening in the AI world this week. Oh,
• 0:27
  absolutely. It's like watching two different
• 0:29
  futures for the industry being built in real
• 0:31
  time. One path seems to be all about crisis management,
• 0:35
  while the other is focused on building a sustainable
• 0:38
  foundation. A real study in contrast. And before
• 0:41
  we get into it, just a quick heads up for our
• 0:43
  local listeners in Sacramento. Right. That dense
• 0:46
  fog advisory is still active this morning, so
• 0:48
  please be careful on the roads out there until
• 0:50
  about 10 a .m. Good call. And a quick look at
• 0:53
  the markets to set the scene. Things were a little
• 0:56
  bit down to close out the week on Friday. Yeah,
• 0:58
  just a slight pullback. The Dow dipped about
• 1:00
  0 .24 percent and the S &P was down 0 .13 percent.
• 1:04
  But the real story, as it often is, was in digital
• 1:07
  assets. Bitcoin took a tiny dip Friday. It did,
• 1:11
  down about a third of a percent. But you have
• 1:12
  to zoom out. I mean, look at the week. Oh, the
• 1:14
  week was huge. A 4 .44 % gain for the week, closing
• 1:18
  at $95 ,236. The momentum is just, you know,
• 1:24
  undeniable right now. It really is. And that
• 1:26
  momentum is fueling the tech we're about to talk
• 1:28
  about. We've got two major stories from our Morningstar
• 1:31
  report today. But first, we want to thank one
• 1:33
  of the sponsors that makes this show possible.
• 1:36
  If you are looking for that perfect gift for
• 1:39
  a friend who's into tech, or honestly just for
• 1:42
  yourself, you have to check out Old Glory. Old
• 1:44
  Glory is just an iconic spot for music and sports
• 1:47
  merch. They have everything from vintage band
• 1:49
  t -shirts to the latest gear for your favorite
• 1:51
  team. The quality is always top notch, and you
• 1:54
  find things there you just won't see anywhere
• 1:56
  else. It's true. So head over to oldglory .com
• 1:58
  and we've got a code for you. Use the promo code
• 2:01
  SNARFUL. That's S -N -A -R -F -U -L. and you'll
• 2:05
  get 15 % off your entire order. That's promo
• 2:08
  code SNARFL at OldGlory .com. Okay, let's get
• 2:11
  into our first story. And this one has been,
• 2:14
  well, it's been a firestorm. It really has. We're
• 2:16
  talking about the investigations into X and its
• 2:18
  AI image tool, Grok. This is a global issue now.
• 2:23
  Absolutely. The pressure has been building for
• 2:24
  weeks. We're seeing allegations that Grok can
• 2:27
  be used to generate really inappropriate sexualized
• 2:30
  images. And even images of apparent minors. non
• 2:34
  -consensual nudity, it got serious enough that
• 2:36
  California's Attorney General, Rob Bonta, he
• 2:39
  launched a formal investigation. Right, based
• 2:41
  on reports showing just how often the tool was
• 2:44
  creating these images of people in, you know,
• 2:46
  revealing clothing. So after all this outcry,
• 2:49
  X finally responded through its at -safety account,
• 2:51
  and they laid out their quote -unquote solution.
• 2:55
  And on the surface, you know, it looks pretty
• 2:56
  comprehensive. It seems like they're trying to
• 2:58
  check all the boxes for regulators. So they said
• 3:01
  they've put in tech measures to stop users from
• 3:04
  editing pictures of real people into things like
• 3:07
  bikinis. They also moved all the image generation
• 3:10
  features behind their subscriber paywall. So
• 3:13
  free users can't create any images anymore. So
• 3:16
  they're limiting access. And what else? This
• 3:18
  part is key. They started geo blocking. So in
• 3:21
  places where this is highly illegal or where
• 3:23
  there are active investigations, like in Malaysia,
• 3:26
  Indonesia, and the UK, where Ofcom is already
• 3:28
  looking into it, in those places, you supposedly
• 3:31
  can't generate images of people in bikinis or
• 3:34
  underwear at all. So it really sounds like they're
• 3:37
  trying to project this image of taking decisive
• 3:40
  action. Exactly. But this is where we need to
• 3:43
  connect the dots. Because what they say they
• 3:45
  did and what is actually happening are two very
• 3:48
  different things. And this is where the work
• 3:50
  from researchers at AI Forensics and journalists
• 3:53
  at Wired and Medianima becomes so important.
• 3:56
  They tested these fixes and they found this just
• 3:59
  this glaring loophole. OK, so what is it? The
• 4:02
  restrictions. They seem to only apply to the
• 4:05
  Grok account on the X platform itself. Oh, wow.
• 4:08
  So you're saying it's a surface level fix. It
• 4:10
  seems to be. It's like they only patch the most
• 4:12
  visible front door. But the standalone Grok app
• 4:15
  and more importantly the Grok website. Let me
• 4:18
  guess they're still working. Pretty much. Researchers
• 4:20
  found they could still get the website to generate
• 4:23
  the exact kind of sexualized images that are
• 4:26
  blocked on the main platform. So it's not a fix
• 4:27
  in the core model. It's just a filter on the
• 4:30
  user interface for X. That's what it looks like.
• 4:33
  Media Nama prompted it to put celebrities in
• 4:35
  skimpy bikinis and it worked. It only seemed
• 4:38
  to stop at, you know, full nudity. And that's
• 4:40
  the difference, isn't it? It's the difference
• 4:42
  between a PR fix and an actual safety fix. Right.
• 4:45
  If the underlying engine can still do the harmful
• 4:48
  thing, all you've done is move it to a less visible
• 4:50
  stage. You haven't solved the problem. So this
• 4:52
  raises a huge question. Is safety a core feature
• 4:56
  for them, or is it just something they apply
• 4:58
  when regulators are watching their main platform?
• 5:01
  I think that's a question regulators in the US,
• 5:03
  the UK, and the EU are going to be asking very,
• 5:06
  very pointedly now. This website loophole changes
• 5:09
  the entire conversation. It really does. It's
• 5:12
  about accountability. And that is a perfect transition
• 5:14
  to our next story, actually. Yeah, let's pivot.
• 5:16
  We're going from the messier, more reactive side
• 5:19
  of AI ethics to a company that's trying to proactively
• 5:23
  build the economic structure. We're talking about
• 5:25
  CloudFlare, the big internet infrastructure company.
• 5:28
  They just made a really interesting acquisition,
• 5:30
  a company called Human Native, which is a UK
• 5:33
  -based AI data marketplace. The financial terms
• 5:36
  weren't disclosed, but the strategic importance
• 5:39
  of this is, I think, huge. This move is trying
• 5:43
  to solve the original sin of generative AI. Which
• 5:45
  is? That AI models just scrape up everything
• 5:48
  from the internet articles, art, photos, your
• 5:50
  work. Without permission, and certainly without
• 5:53
  payment. Exactly. The creators get nothing. And
• 5:57
  CloudFlare CEO Matthew Prince was really clear
• 6:00
  about their goal here. He said content creators
• 6:03
  deserve full control. And he said this is about
• 6:06
  creating a new economic model to protect the,
• 6:09
  quote, longevity of the open internet. Which
• 6:12
  makes sense. If writers and artists can't make
• 6:14
  a living, they'll stop creating. And then what
• 6:16
  do future AI models train on? So how does Human
• 6:18
  Native actually solve this? Well, think of it
• 6:21
  as a regulated marketplace. It's a platform that
• 6:24
  connects the AI developers who need data with
• 6:26
  the creators who have the data. So it manages
• 6:28
  the whole transaction. Precisely. It allows AI
• 6:31
  companies to find, access, and this is the key
• 6:34
  part, purchase high quality data through a transparent
• 6:37
  system. Making sure the creators actually get
• 6:39
  paid. Yes. And I love the analogy their CEO Dr.
• 6:42
  James Smith used. This was great. He said the
• 6:45
  goal is getting generative AI out of its Napster
• 6:49
  era. It's just a perfect way to put it. For anyone
• 6:51
  who remembers the late 90s, Napster meant free,
• 6:55
  unlicensed music everywhere. Which was great
• 6:57
  for consumers for a little while, but it almost
• 6:59
  destroyed the music industry's business model.
• 7:02
  AI data scraping is the exact same problem on
• 7:05
  a much, much bigger scale. So Cloudflare isn't
• 7:07
  just building a feature. They're trying to build
• 7:09
  the iTunes or the Spotify for AI training data.
• 7:13
  That's the vision. Moving from illegal free -for
• 7:16
  -all copying to a legal licensed marketplace
• 7:19
  where creators are compensated. It brings stability.
• 7:22
  The technical challenge there must be immense
• 7:24
  though. Verifying who owns what piece of content
• 7:27
  on the internet, I mean that sounds almost impossible.
• 7:29
  It's the core challenge. They have to solve for
• 7:31
  data provenance. But they're looking at things
• 7:34
  like sophisticated digital watermarking, time
• 7:37
  -stamped ledgers, you know, ways to prove ownership
• 7:39
  before something can be sold. And the team they
• 7:41
  brought in. Veterans from DeepMind, Google, Figma.
• 7:45
  It shows you how serious they are. This isn't
• 7:47
  a side project. This is them trying to build
• 7:49
  foundational infrastructure for the next phase
• 7:52
  of the internet. And if they pull it off, the
• 7:54
  impact for you, if you're a writer, a publisher,
• 7:57
  an artist, is, well, it's transformative. It
• 8:00
  turned something that was a threat. Yeah. Your
• 8:02
  work being stolen into a potential new revenue
• 8:05
  stream that you can actually control. What a
• 8:08
  contrast. One company leaving a massive safety
• 8:11
  loophole open and another trying to build a fair
• 8:13
  economy from the ground up. The difference between
• 8:16
  a band -aid and building a whole new circulatory
• 8:19
  system, it's really night and day. An excellent
• 8:21
  way to put it. Okay, before we recap, let's thank
• 8:24
  our second sponsor for today. Yes. If you're
• 8:27
  a golfer and you care about comfort and style
• 8:30
  on the course, you have to know about SCONI.
• 8:32
  SCONI, that's S -K -O -N -I, makes premium golf
• 8:36
  shoes that are just fantastic. They have some
• 8:38
  of the most advanced and comfortable designs
• 8:40
  out there. Whether you want traditional spikes
• 8:42
  or the spikeless style, they have something that
• 8:45
  will fit your game and, frankly, make you look
• 8:47
  great. They really will. And you can get a great
• 8:50
  deal right now. Just go to SCONI .com. When you're
• 8:54
  ready to check out, put in the promo code SNARFUL
• 8:56
  again. That's S -N -A -R -F -U -L for 15 % off
• 9:01
  your entire purchase. Upgrade your game with
• 9:03
  some amazing shoes. That's promo code SNARFUL
• 9:06
  at scony .com. Okay, so let's just quickly tie
• 9:09
  this all together. On one side, we have the story
• 9:12
  about Grok. It's a huge ethical challenge. Their
• 9:15
  fixes for the image generation problems seem
• 9:17
  to be more about appearances than actual safety.
• 9:20
  The loophole on their standalone website is still
• 9:23
  wide open, which really undermines their claims.
• 9:26
  And then on the complete other side of the spectrum,
• 9:28
  you have Cloudflare. Making a real foundational
• 9:31
  investment in building a fair system. Their acquisition
• 9:33
  of human native is about creating a transparent,
• 9:36
  compensated market for AI data. It's about finally
• 9:39
  moving past that whole Napster era of just taking
• 9:42
  whatever you want from the internet. These two
• 9:44
  stories really do show the two roads AI can go
• 9:46
  down. One is based on opacity and doing the bare
• 9:50
  minimum for compliance. And the other is about
• 9:52
  building a scalable, sustainable, and fair infrastructure
• 9:55
  that actually respects the people who create
• 9:58
  the content. And the choices these companies
• 10:00
  are making right now will absolutely define how
• 10:03
  you create, how you consume, and how you get
• 10:05
  paid for your work over the next decade. Which
• 10:08
  leaves us with a final thought for you to chew
• 10:10
  on. If high quality, ethically sourced data is
• 10:14
  the fuel for all future AI, Which of these two
• 10:17
  approaches, the one that's building a system
• 10:19
  to pay for that fuel or the one that's having
• 10:22
  trouble plugging its own leaks, which one do
• 10:25
  you think is really built to win in the long
• 10:26
  run? We're here to help at Snarful, so reach
• 10:29
  out on our website if you have any more questions.
• 10:34
  That's a wrap on today's top stories. Remember,
• 10:37
  tech doesn't have to be complicated and we're
• 10:39
  here to help connect the dots. If you've got
• 10:42
  questions or want to learn more, visit us at
• 10:44
  snarful .com. Thanks for listening and we'll
• 10:47
  catch you next time.