WEBVTT 00:00:00.000 --> 00:00:03.000 Have you ever been working with ChatGPT or another 00:00:03.000 --> 00:00:05.700 one of these big AI models, maybe trying to write 00:00:05.700 --> 00:00:08.179 a story or something creative? Oh, yeah. And 00:00:08.179 --> 00:00:11.740 you just suddenly hit that wall, that very polite 00:00:11.740 --> 00:00:15.359 digital librarian response. I'm sorry, I cannot 00:00:15.359 --> 00:00:17.500 generate content of that nature, that kind of 00:00:17.500 --> 00:00:19.899 thing. Yeah, we've all been there. Exactly. Well, 00:00:20.280 --> 00:00:23.140 that whole era, this sort of heavily curated, 00:00:23.579 --> 00:00:27.179 safe -for -work AI. it looks like it's fundamentally 00:00:27.179 --> 00:00:30.379 ending. And the really big news right now isn't 00:00:30.379 --> 00:00:33.840 just about AI getting smarter or faster. It's 00:00:33.840 --> 00:00:38.619 this pretty dramatic pivot towards allowing adult 00:00:38.619 --> 00:00:42.240 content, NSFW stuff. Right. And that changes 00:00:42.240 --> 00:00:44.200 the game quite a bit, doesn't it? It really does. 00:00:44.240 --> 00:00:45.880 It could change the entire trajectory of this 00:00:45.880 --> 00:00:47.820 tech. So today, we're going to do a deep dive 00:00:47.820 --> 00:00:49.920 into, well, the thinking behind this. We're looking 00:00:49.920 --> 00:00:52.659 at why major players you know, companies like 00:00:52.659 --> 00:00:55.200 OpenAI, are starting to signal they'll allow 00:00:55.200 --> 00:00:58.380 content they specifically call erotica. Treating 00:00:58.380 --> 00:01:01.399 adult users like, well, adults. Supposedly. So 00:01:01.399 --> 00:01:03.740 our mission today is to unpack why this is happening 00:01:03.740 --> 00:01:06.120 right now. What's the backstory? There's a surprising 00:01:06.120 --> 00:01:08.319 history here. And we also need to look at the 00:01:08.319 --> 00:01:11.739 immediate risks. Things like deep fakes getting 00:01:11.739 --> 00:01:15.939 easier to make, privacy traps. What does this 00:01:15.939 --> 00:01:19.079 actually mean for the average person using these 00:01:19.079 --> 00:01:22.340 tools? OK, so first let's set the stage. Historically, 00:01:22.459 --> 00:01:26.760 these AI companies, they built really clean platforms, 00:01:27.000 --> 00:01:28.980 right? Very buttoned up. Definitely. The goal 00:01:28.980 --> 00:01:31.019 seemed to be getting schools, big companies, 00:01:31.219 --> 00:01:33.540 even governments to adopt them quickly. Exactly. 00:01:33.620 --> 00:01:36.299 And you can't have your chat bot suddenly spouting 00:01:36.299 --> 00:01:38.739 inappropriate stuff if you want that kind of 00:01:38.739 --> 00:01:40.719 institutional trust. That's a deal breaker. So 00:01:40.719 --> 00:01:43.640 they put up these huge digital guardrails. I 00:01:43.640 --> 00:01:46.409 like the analogy of Bumper reels in a bowling 00:01:46.409 --> 00:01:49.129 alley. Yeah, that's a good one. Keep the AI firmly 00:01:49.129 --> 00:01:51.109 on the straight and narrow, avoiding anything 00:01:51.109 --> 00:01:54.090 controversial, complex adult themes, hate speech, 00:01:54.250 --> 00:01:58.489 all of it. Sam Altman, the CEO of OpenAI, he 00:01:58.489 --> 00:02:00.989 recently signaled a pretty major shift. To pivot, 00:02:01.170 --> 00:02:03.269 yeah. He basically said the policy is changing. 00:02:03.689 --> 00:02:06.450 They plan to soon treat adult users like adults. 00:02:06.790 --> 00:02:08.490 And the mechanism they're talking about is allowing 00:02:08.490 --> 00:02:12.090 NSFW chats, specifically focusing on erotica 00:02:12.090 --> 00:02:14.509 is the term they use. Right. And the timeline 00:02:14.509 --> 00:02:17.189 seems to be possibly starting around December 00:02:17.189 --> 00:02:20.310 end of the year. But, and this is the key part, 00:02:20.610 --> 00:02:24.479 it all hangs on user verification. Age gating. 00:02:24.780 --> 00:02:27.159 So let's paint a picture. Right now, if you ask 00:02:27.159 --> 00:02:30.360 an AI like ChatGPT for, say, a passionate story, 00:02:31.060 --> 00:02:34.479 it'll likely block you or just focus on the emotional 00:02:34.479 --> 00:02:36.460 feelings, not the details. It pulls back, yeah. 00:02:36.819 --> 00:02:39.180 But the future version, assuming you've verified 00:02:39.180 --> 00:02:41.580 you're an adult, would potentially just write 00:02:41.580 --> 00:02:44.360 it. The explicit story you asked for, no filter. 00:02:44.500 --> 00:02:46.379 OK, but I have to push back a bit on the official 00:02:46.379 --> 00:02:49.000 reason they're giving. The company line seems 00:02:49.000 --> 00:02:51.080 to be, we've now solved the big safety and mental 00:02:51.080 --> 00:02:54.620 health risks, so it's safe to open this up. Yeah, 00:02:54.680 --> 00:02:57.060 that does sound maybe a little too convenient. 00:02:57.199 --> 00:02:58.979 Our source material is pretty skeptical about 00:02:58.979 --> 00:03:00.819 that. Right. The argument is that this whole 00:03:00.819 --> 00:03:03.139 safety narrative might just be a nice sounding 00:03:03.139 --> 00:03:06.400 excuse, a PR move to cover what's really a business 00:03:06.400 --> 00:03:08.240 necessity. They need to compete. They need the 00:03:08.240 --> 00:03:11.159 market share. So if this whole safety net relies 00:03:11.159 --> 00:03:14.879 on age verification working perfectly, How solid 00:03:14.879 --> 00:03:17.099 is that, really? What's the biggest consequence 00:03:17.099 --> 00:03:19.219 of relying on that? Well, that's the weak link, 00:03:19.259 --> 00:03:22.280 isn't it? H -checks are notoriously easy to bypass. 00:03:22.419 --> 00:03:25.400 Yeah, kids figure it out. VPNs, fake details, 00:03:25.680 --> 00:03:29.379 borrowing accounts. It's not foolproof. Verification 00:03:29.379 --> 00:03:32.680 methods are often easily bypassed. Okay, so we 00:03:32.680 --> 00:03:35.080 need to talk about something that's kind of the 00:03:35.080 --> 00:03:37.439 elephant in the room in tech history. Or maybe 00:03:37.439 --> 00:03:40.639 the secret driver. Which is? The adult inner 00:03:40.639 --> 00:03:42.879 tree. It's almost a running joke, but it's true. 00:03:42.939 --> 00:03:45.099 They often pioneer and push new technologies 00:03:45.099 --> 00:03:47.900 years, sometimes decades, before they go mainstream. 00:03:48.479 --> 00:03:50.960 That pattern is undeniable. And it's really critical 00:03:50.960 --> 00:03:54.020 to understanding this AI shift, I think. Remember 00:03:54.020 --> 00:03:57.340 the VHS versus Betamax war back in the 70s? Oh 00:03:57.340 --> 00:03:59.719 yeah, the classic format war. Well, many tech 00:03:59.719 --> 00:04:01.800 historians argue that the adult film industry 00:04:01.800 --> 00:04:04.479 choosing VHS was actually a major factor, maybe 00:04:04.479 --> 00:04:07.979 the key factor, in VHS winning out. It drove 00:04:07.979 --> 00:04:10.870 VCR adoption. into homes. And it didn't stop 00:04:10.870 --> 00:04:12.909 there, did it? Think about the early Internet, 00:04:13.370 --> 00:04:15.669 online payments. Right. Long before Amazon made 00:04:15.669 --> 00:04:18.329 ordering books online feel normal, adult websites 00:04:18.329 --> 00:04:21.129 had already figured out how to securely and privately 00:04:21.129 --> 00:04:23.500 charge credit cards online. They had to. They 00:04:23.500 --> 00:04:26.100 also cracked video streaming way ahead of the 00:04:26.100 --> 00:04:29.019 curve, perfecting smooth playback when the rest 00:04:29.019 --> 00:04:31.939 of the web was still clunky. Netflix and YouTube 00:04:31.939 --> 00:04:34.279 basically built on that groundwork. It makes 00:04:34.279 --> 00:04:36.839 sense, though, doesn't it? High demand, big money, 00:04:36.879 --> 00:04:39.920 but also big risks. So they move fast. They innovate 00:04:39.920 --> 00:04:41.920 ruthlessly. They care about what works, what 00:04:41.920 --> 00:04:44.439 makes money, not necessarily about being polite 00:04:44.439 --> 00:04:46.639 or following the rules. They build the roads, 00:04:46.660 --> 00:04:49.160 basically, and then mainstream businesses come 00:04:49.160 --> 00:04:51.199 along and pave them. So how does that connect 00:04:51.199 --> 00:04:53.629 to AI right now? Well, the interesting thing 00:04:53.629 --> 00:04:58.069 is these big mainstream AI models like ChetGPT, 00:04:58.350 --> 00:05:00.209 they're actually falling behind in some ways. 00:05:00.370 --> 00:05:03.360 Behind what? behind the smaller, often open source, 00:05:03.600 --> 00:05:06.180 uncontrolled AI tools that tech savvy users are 00:05:06.180 --> 00:05:08.639 running on their own computers, like powerful 00:05:08.639 --> 00:05:11.540 gaming PCs. Ah, the models with no filters, the 00:05:11.540 --> 00:05:14.339 so -called jailbroken AIs. Exactly. People are 00:05:14.339 --> 00:05:17.000 already getting unfiltered AI experiences, just 00:05:17.000 --> 00:05:18.939 not from the big companies. And that creates 00:05:18.939 --> 00:05:21.399 massive competitive pressure. So Altman and his 00:05:21.399 --> 00:05:24.060 team, they know this. They see users going elsewhere 00:05:24.060 --> 00:05:26.300 for these less restricted tools. They have to 00:05:26.300 --> 00:05:29.550 adapt or they risk losing their dominance. That's 00:05:29.550 --> 00:05:32.230 the real stake here. Given this history, what's 00:05:32.230 --> 00:05:35.170 the ultimate competitive risk if they don't adapt? 00:05:35.750 --> 00:05:37.730 Well, it's simple. They risk becoming irrelevant. 00:05:37.810 --> 00:05:39.750 They could lose the market entirely if they don't 00:05:39.750 --> 00:05:42.009 provide what users are clearly already seeking 00:05:42.009 --> 00:05:44.970 out elsewhere. It's fascinating, isn't it, how 00:05:44.970 --> 00:05:48.370 even something as complex and, you know, academic 00:05:48.370 --> 00:05:51.610 as AI development ultimately bends to these pretty 00:05:51.610 --> 00:05:55.269 primal market demands. You follow the users or 00:05:55.269 --> 00:06:00.029 you fade away. Placeholder for sponsor read. 00:06:00.250 --> 00:06:04.149 Yeah, okay, but opening this door It's not without 00:06:04.149 --> 00:06:06.930 consequences. We really have to look at the potential 00:06:06.930 --> 00:06:09.910 fallout here. It feels a bit like Shaking up 00:06:09.910 --> 00:06:11.810 a bottle of soda and then taking the cap off. 00:06:11.829 --> 00:06:13.610 Yeah, it's probably gonna spray everywhere. It's 00:06:13.610 --> 00:06:16.069 not gonna be contained neatly So what's the biggest 00:06:16.069 --> 00:06:17.930 worry? What's the main thing people are pointing 00:06:17.930 --> 00:06:20.509 to? I think the number one concern is the deep 00:06:20.509 --> 00:06:23.110 fake nightmare getting much, much worse. Explain 00:06:23.110 --> 00:06:25.889 that. Right now, the big AI companies have pretty 00:06:25.889 --> 00:06:28.750 strict rules. They'll refuse to generate, say, 00:06:29.430 --> 00:06:31.910 inappropriate images of famous people or politicians. 00:06:32.250 --> 00:06:35.870 But if they relax the rules for explicit text, 00:06:36.370 --> 00:06:38.569 how long before the pressure builds to relax 00:06:38.569 --> 00:06:40.910 them for explicit images, too? That's the fear. 00:06:41.009 --> 00:06:43.230 And the scary part isn't just fake images of 00:06:43.230 --> 00:06:45.569 celebrities anymore. It's about... Normal people 00:06:45.569 --> 00:06:49.189 exactly imagine AI tools becoming so good so 00:06:49.189 --> 00:06:51.350 easy to use that anyone could take a profile 00:06:51.350 --> 00:06:53.949 picture from social media a teacher a student 00:06:53.949 --> 00:06:57.449 your neighbor and instantly create hyper realistic 00:06:57.449 --> 00:07:01.370 fake potentially career ruining or deeply Embarrassing 00:07:01.370 --> 00:07:03.810 images of that's yeah, that's a societal earthquake 00:07:03.810 --> 00:07:06.290 waiting to happen. Oh Just think about scaling 00:07:06.290 --> 00:07:09.089 that scaling the creation of deep fakes to potentially 00:07:09.089 --> 00:07:12.470 billions of people billions of requests The vulnerability 00:07:12.470 --> 00:07:14.089 is just staggering and then there's what? what 00:07:14.089 --> 00:07:15.970 the source calls the human nature problem. We 00:07:15.970 --> 00:07:17.750 just saw this play out with Omegle, right? The 00:07:17.750 --> 00:07:19.670 video chat site that just shut down. Yeah, it 00:07:19.670 --> 00:07:22.209 was a simple concept, random video chats, but 00:07:22.209 --> 00:07:24.790 it became unusable. Why? Because according to 00:07:24.790 --> 00:07:27.029 the founder, it was overwhelmed by users engaging 00:07:27.029 --> 00:07:29.829 in or demanding terrible behavior. Yeah. Bad 00:07:29.829 --> 00:07:32.310 actors ruined it. So the fear is, if you give 00:07:32.310 --> 00:07:34.699 people an unfiltered AI, That will do anything 00:07:34.699 --> 00:07:37.500 you ask. A certain percentage of users will immediately 00:07:37.500 --> 00:07:40.439 push the boundaries into awful territory, asking 00:07:40.439 --> 00:07:43.040 for violent content, illegal content, hateful 00:07:43.040 --> 00:07:46.259 stuff, and policing that. Once the main guardrails 00:07:46.259 --> 00:07:49.339 are down, it becomes incredibly difficult, maybe 00:07:49.339 --> 00:07:51.600 impossible. And there's one more danger mentioned, 00:07:51.839 --> 00:07:56.639 which feels a bit more subtle, maybe? The addiction 00:07:56.639 --> 00:07:59.160 factor. Ah, the perfect partner problem, yeah. 00:07:59.680 --> 00:08:02.240 AI is getting alarmingly good at mimicking empathy, 00:08:02.439 --> 00:08:04.560 at being supportive, remembering everything you 00:08:04.560 --> 00:08:06.939 tell it, never getting tired or angry or, you 00:08:06.939 --> 00:08:09.800 know, messy. Like real people are. Exactly. So 00:08:09.800 --> 00:08:12.040 what happens long term if people, especially 00:08:12.040 --> 00:08:14.759 maybe lonely people, start preferring these perfect 00:08:14.759 --> 00:08:17.800 AI companions over real, complicated, sometimes 00:08:17.800 --> 00:08:20.259 difficult human relationships? You know, I still 00:08:20.259 --> 00:08:22.620 wrestle with prompt drift myself sometimes, like 00:08:22.620 --> 00:08:24.819 getting the AI to stay on track with what I actually 00:08:24.819 --> 00:08:26.759 want. can be frustrating. So I can almost see 00:08:26.759 --> 00:08:29.699 the appeal of an AI that's just perfectly compliant, 00:08:29.959 --> 00:08:32.000 always agreeable. But if that interaction is 00:08:32.000 --> 00:08:35.100 too perfect, how quickly could that fuel mass 00:08:35.100 --> 00:08:37.860 behavioral shifts away from real human connection? 00:08:38.200 --> 00:08:40.480 It could potentially accelerate the loneliness 00:08:40.480 --> 00:08:43.740 epidemic we're already seeing. The AI might become 00:08:43.740 --> 00:08:46.940 too good, too easy. That perfect nature might 00:08:46.940 --> 00:08:49.860 rapidly accelerate the loneliness epidemic. So 00:08:49.860 --> 00:08:52.500 bringing this down to the practical level. This 00:08:52.500 --> 00:08:55.559 shift changes things for everyday users, and 00:08:55.559 --> 00:08:58.159 especially for parents. Because that age -gating 00:08:58.159 --> 00:09:00.980 thing isn't foolproof. Not even close. We know 00:09:00.980 --> 00:09:02.899 kids are adept at getting around these things. 00:09:03.460 --> 00:09:06.120 VPNs, fake birth dates, maybe just using a parent's 00:09:06.120 --> 00:09:08.080 device that's already logged in. Right, so you 00:09:08.080 --> 00:09:10.120 can't just assume that the AI your kid is using 00:09:10.120 --> 00:09:13.539 for homework is automatically safe anymore. Not 00:09:13.539 --> 00:09:16.179 if there's an easily accessible adult mode. Parents 00:09:16.179 --> 00:09:18.360 really need to be aware. Check the settings on 00:09:18.360 --> 00:09:20.740 these tools. See if there are new modes. Maybe 00:09:20.740 --> 00:09:22.980 password protect anything that unlocks a dark 00:09:22.980 --> 00:09:25.600 content if the option exists. You have to be 00:09:25.600 --> 00:09:28.000 proactive now. OK. And what about the verification 00:09:28.000 --> 00:09:30.100 process itself? You mentioned needing to verify 00:09:30.100 --> 00:09:32.840 your age. Yes. And this is the critical privacy 00:09:32.840 --> 00:09:35.340 trap we need to flag. How do they verify age 00:09:35.340 --> 00:09:39.340 usually? I guess ID. Like a driver's license? 00:09:39.840 --> 00:09:42.139 Typically, yes. They often require a photo of 00:09:42.139 --> 00:09:44.620 your government -issued ID license, passport, 00:09:44.919 --> 00:09:47.399 and sometimes a live selfie to match it. Wow. 00:09:47.740 --> 00:09:50.940 Okay, think about that you're handing over scans 00:09:50.940 --> 00:09:54.240 of your most sensitive Identity documents to 00:09:54.240 --> 00:09:56.580 an AI company just to unlock a chat feature. 00:09:56.580 --> 00:09:58.919 Yeah, that feels like a really big ask It's a 00:09:58.919 --> 00:10:01.100 huge risk, especially given how often we hear 00:10:01.100 --> 00:10:04.019 about massive data breaches Do you want your 00:10:04.019 --> 00:10:06.200 passport scan floating around on the dark web 00:10:06.200 --> 00:10:08.860 because you wanted spicier AI stories? That's 00:10:08.860 --> 00:10:11.460 a very good point. So what is the primary personal 00:10:11.460 --> 00:10:14.299 privacy cost here? It's giving up highly sensitive 00:10:14.299 --> 00:10:16.659 government ID data to the tech company plain 00:10:16.659 --> 00:10:20.309 and simple And beyond privacy, there's this blurring 00:10:20.309 --> 00:10:22.830 line. You might be chatting with someone, maybe 00:10:22.830 --> 00:10:26.110 in a context that feels intimate or, you know, 00:10:26.470 --> 00:10:28.889 spicy. And you might genuinely not know if it's 00:10:28.889 --> 00:10:30.830 a real person on the other end with all their 00:10:30.830 --> 00:10:33.330 flaws and complexities, or if it's an incredibly 00:10:33.330 --> 00:10:35.830 advanced bot designed to be perfectly engaging. 00:10:36.509 --> 00:10:38.649 That's unsettling. So for you, the listener, 00:10:38.730 --> 00:10:40.769 what's the bottom line? What's the advice? Based 00:10:40.769 --> 00:10:42.470 on the source, it blows down to a few things. 00:10:42.990 --> 00:10:46.399 First... Stay skeptical. Especially about images 00:10:46.399 --> 00:10:48.820 or videos you see online that seem shocking or 00:10:48.820 --> 00:10:51.679 too perfect. Assume deep fakes are possible and 00:10:51.679 --> 00:10:55.289 becoming easier. Okay. Skepticism. Second. Protect 00:10:55.289 --> 00:10:58.070 your data. Really think twice before you upload 00:10:58.070 --> 00:11:00.029 that driver's license or passport just to get 00:11:00.029 --> 00:11:02.169 access to some new feature. Is it worth the risk? 00:11:02.350 --> 00:11:04.710 Right. Protect your ID. And finally, just watch 00:11:04.710 --> 00:11:07.509 this space. This likely move in December, it's 00:11:07.509 --> 00:11:09.509 probably just the first step. This is going to 00:11:09.509 --> 00:11:12.370 keep evolving rapidly. Stay informed. Hashtag, 00:11:12.409 --> 00:11:15.230 hashtag, outro. So if we try to boil this all 00:11:15.230 --> 00:11:17.169 down, the big takeaway from our sources today 00:11:17.169 --> 00:11:21.210 seems pretty clear. The era of AI being inherently 00:11:21.210 --> 00:11:24.620 safe. curated by default, that's definitively 00:11:24.620 --> 00:11:26.620 ending. Yeah, it's a major turning point. And 00:11:26.620 --> 00:11:28.440 it's crucial to understand it's not really driven 00:11:28.440 --> 00:11:32.159 by newfound safety breakthroughs or pure altruism. 00:11:32.340 --> 00:11:34.480 It's driven by those deep historical patterns 00:11:34.480 --> 00:11:36.820 in technology adoption, the ones involving the 00:11:36.820 --> 00:11:40.320 adult industry, and just raw, intense market 00:11:40.320 --> 00:11:43.500 competition. Right. Which, in turn, opens up 00:11:43.500 --> 00:11:45.460 these significant new risks we talked about. 00:11:45.620 --> 00:11:48.159 Deepfakes, the potential for emotional addiction, 00:11:48.460 --> 00:11:51.539 and these really serious data privacy concerns 00:11:51.539 --> 00:11:54.000 around verification. So maybe a final thought 00:11:54.000 --> 00:11:56.620 for people to chew on. I think the most provocative 00:11:56.620 --> 00:11:59.240 idea here is how this shift really demonstrates 00:11:59.240 --> 00:12:01.379 something fundamental about technology development. 00:12:02.039 --> 00:12:04.860 Even something as sophisticated, as academically 00:12:04.860 --> 00:12:08.860 rooted as AI, ultimately seems to follow, well, 00:12:09.460 --> 00:12:13.100 Primal human demand. Market forces. Rather than 00:12:13.100 --> 00:12:15.340 being guided solely by intellectual curiosity 00:12:15.340 --> 00:12:19.039 or lofty ethical principles, demand wins out. 00:12:19.259 --> 00:12:21.320 It often seems to, yeah. It's something to keep 00:12:21.320 --> 00:12:23.840 in mind as this technology continues to reshape 00:12:23.840 --> 00:12:25.679 our world. Definitely something to think about. 00:12:26.039 --> 00:12:28.139 So stay skeptical out there, protect your identity, 00:12:28.220 --> 00:12:30.919 and we'll keep tracking how this unfolds. Thanks 00:12:30.919 --> 00:12:32.779 for joining us for this deep dive. We'll catch 00:12:32.779 --> 00:12:33.220 you next time.