BreakDecks
@BreakDecks@lemmy.ml
- Comment on Instagram Advertises Nonconsensual AI Nude Apps 6 months ago:
The pedos out there are using AI to nudity pictures of real kids. That’s just going to drive up the demand for creep shots and child model photosets to exploit.
There may be a small percentage of offending pedophiles that switch to pure GenAI over pictures of real kids, but I don’t see GenAI ever playing a role in harm reduction given the harm it ultimately enables.
One of the current sickening trends is for a predator to convince a kid to send underwear or swimsuit pics, and then blackmail them into more hardcore photos with nudified versions of the original pics. They’re already seeing an influx of that kind of CSAM online, that involves abusing real kids on social media.
I just wish America was less puritanical and taught kids about sex and boundaries to protect them, and that we had a functioning mental healthcare system that directly helps people who experience inappropriate sexuality attractions like pedophilia before they go down these dark paths.
- Comment on Paedophiles create nude AI images of children to extort from them, says charity | Internet safety | The Guardian 6 months ago:
I want to be happy that the IWF exists and is collecting data about this kind of thing. This is extremely difficult and important work.
But they are also lobbying to ban encryption, arguing that privacy only helps criminals.
Sorry, but if Facebook is too dangerous for kids, instead of banning encryption so the authorities can more easily catch child abusers, let’s just ban children from using Facebook before they fall victim to abuse.
- Comment on Instagram Advertises Nonconsensual AI Nude Apps 6 months ago:
That’s kinda why I brought up specific key players and now I consider them complicit. If you don’t want AI to be blamed as a whole, you should want those key players to behave ethically, or they’ll poison public perception of AI as a whole.
- Comment on It must confuse English learners to hear phrases like, "I'm home", instead of "I am at home." We don't say I'm school, or I'm post office. 6 months ago:
My favorite confusing English sentence is “I have had too much to eat.”
- “Have had” is the same word twice, once in present tense, and again in past tense. It counts as one verb.
- Both “too” and “to” used.
- “Eat” is a noun.
- Submitted 6 months ago to showerthoughts@lemmy.world | 42 comments
- Comment on Instagram Advertises Nonconsensual AI Nude Apps 6 months ago:
Generative AI is being used quite prominently for the purposes of making nonconsensual pornography. Just look at the state of CivitAI, the largest marketplace of Stable Diffusion models online. It pretends to be a community for Machine Learning professionals, but behind the scenes it’s laying the groundwork for all of the problems we’re seeing right now. There’s not an actress or female celebrity that doesn’t have a TI or LoRA trained on their likeness - and the galleries don’t hold back on showing you what these models can do.
At least Photoshop never gained the specific reputation of being a tool for making fake porn, but the GenAI community is leaving no doubt that this is a major use case for image models.
Even HuggingFace turns a blind eye to pornifying models and lolicon datasets, and they’re basically the GitHub of AI models…
- Comment on Instagram Advertises Nonconsensual AI Nude Apps 6 months ago:
I logged into my throwaway account today just to check in on it since people are talking about this shit more. I was immediately greeted with an ad featuring hardcore pornography, among the pics of kids that still populate my feed.
I’ll spare you the screenshot, but IG is fucked.
- Comment on Instagram Advertises Nonconsensual AI Nude Apps 6 months ago:
This 100%. I can’t even bring myself to buy new content for my Quest now that I’m aware of the issues (no matter how much I want the latest Beat Saber and Synth Riders DLC), especially since Meta’s Horizon, in my experience, puts adults into direct contact with children. At first I just dismissed metaverse games like VRChat or Horizon as being too popular with kids for me to enjoy it, but now I realize that it put me, an adult, straight into voice chats with tweens, which people should fucking know better than to do. My first thought was to log off because I wasn’t having fun in a kid-dominated space, but I have no doubt that these apps are crawling with creeps who see that as a feature rather than a problem.
We need education for parents that sharing pictures of their kids online comes with real risks, as does giving kids free reign to use the Internet. The laissez faire attitude many people have towards social media needs to be corrected, because real harm is already being done.
Most of the parents that post untoward pics of their kids online are chasing down opportunities for their kids to model, and they’re ignoring the fact that a significant volume of engagement these photos receive comes from people objectifying children. There seems to be a pattern that the most revealing outfits get the most engagement, and so future pictures are equally if not more revealing to chase more engagement…
Parents might not understand how disturbing these patterns are until they’ve already dumped thousands of pictures online, and at that point they’re likely to be in denial about what they’re exposing their kids to, and/or too invested to want to reverse course.
We also need to have a larger conversation, as a society, about using kids as models at all. Pretty much every major manufacturer of children’s clothing is hiring real kids to model the clothes. I don’t think it’s necessary to be publishing that many pictures of kids online, nor is it acceptable to be doing so for profit. There’s no reason not to limit modeling to adults who can consent to putting their bodies on public display, and using mannequins for kids’ clothing. The sheer volume of kids’ swimsuit and underwear pictures hosted on e-commerce sites is likely a contributor to the capability Generative AI models have to create inappropriate images of children, not to mention the actual CSAM found in the LAION dataset most of these models are trained on.
Sorry for the long rant, this shit pisses me off. I need to consider sending 404 Media everything I know since they’re doing investigations into this kind of thing. My small scale investigation has revealed a lot to me, but more people need to be getting as upset as I am about it if we want to make the Internet less of a hellscape.
- Comment on Instagram Advertises Nonconsensual AI Nude Apps 6 months ago:
Send 'em to Zuck.
- Comment on Instagram Advertises Nonconsensual AI Nude Apps 6 months ago:
This was from a test I did with a throwaway account on IG where I followed a handful of weirdo parents who run “model” accounts for their kids to see if Instagram would start pushing problematic content as a result (spoiler: yes they will).
It took about 5 minutes from creating the account to end up with nothing but dressed down kids on my recommendations page paired with inappropriate ads. I guess the people who follow kids on IG also like these recommended photos, and the algorithm also figures they must be perverts, but doesn’t care about the sickening juxtaposition of children in swimsuits next to AI nudifying apps.
Don’t use Meta products. They don’t care about ethics, just profits.
- Comment on Instagram Advertises Nonconsensual AI Nude Apps 6 months ago:
Sharing this screenshot again, to drive the point home.
- Comment on Instagram Advertises Nonconsensual AI Nude Apps 6 months ago:
You didn’t have to come out and tell everyone that you’re one of those guys who doesn’t understand the concept of sexual exploitation and consent.
It literally doesn’t matter what you call this. Colloquially the technology is known as “Generative AI”, and it is fully automating the task of making fake nudes to the point that shady websites only require a single input image, and with a few layers of machine learning, are able to spit out a convincing nude.
It was just as fucked up when perverts sexually exploited people with Photoshop, so I don’t understand what your point here is. “AI” has made sexual exploitation fully automated, and there’s absolutely no excuse for defending this.
- Comment on Fisker now expects to go bankrupt within 30 days 6 months ago:
Disposable vapes put more lithium into landfills than EVs. Everyone throws their vape in the trash, nobody throws their EV battery module in the trash
- Comment on Microsoft wants to hide the 'Sign out' button in Windows 11 behind a Microsoft 365 ad 6 months ago:
The Quest is Android based, which is Linux based, so kinda if you’re cool with two layers of megacorp between you and the kernel.
- Comment on Whistleblower 'would not' put family on Boeing 787 jet 7 months ago:
But without planes and their deadly propellers, where will we push Nazis?
- Comment on Whistleblower 'would not' put family on Boeing 787 jet 7 months ago:
Boeing isn’t an airline…
- Comment on Big Tech Is Faking AI 7 months ago:
I submitted a few weird requests to mturk just to see how it works. I was able to read a bunch of magazines for cheap by paying people $0.01 for every two scanned pages of any magazine that was no older than 3 editions old.
I ended up with a ton of random digitized magazines, and ended up learning a lot about the kinds of people who do mturk tasks from the magazines they scanned. Seems to mostly be bored housewives, at least 10 years ago when I did this.
I paid for my experiments with the ~$50 I earned from doing mturk tasks myself, and let me tell you, it was miserable stuff. Sub-minimum wage drudgery… At least I suffered myself what I made others suffer with my stupid tasks, and all I got out of it was a bunch of articles I didn’t actually want to read.
- Comment on Big Tech Is Faking AI 7 months ago:
Anything can be an AI if the marketing department decides to call it AI.
- Comment on ISPs can charge extra for fast gaming under FCC’s Internet rules, critics say 7 months ago:
I have a hard time believing that since TikTok is such a popular mobile app, and 4G users are going to have a minimum of 200ms latency over-the-air.
5G does offer much faster over-the-air latency of only 1ms, but if TikTok were to become unusable if the latency between tower and server exceeds 1ms, I would expect to hear widespread complaints about that, as the vast majority of users aren’t going to have <2ms latency between them and the server. Pinging TikTok.com from my 5G handset (in a major metro area) shows a latency of 80ms. Admittedly, AWS and GCP, where TikTok hosts content primarily for US users, likely has better latency. My personal experience with those platforms doesn’t suggest that it would be that good- pinging the cloudfront CDN endpoints associated with my account, I get latency between 40-60ms.
Besides, video streaming shouldn’t be affected by latency, only throughput. TikTok videos are just MP4 files on a CDN, as long as you can download them faster than you can play them, high latency shouldn’t even be noticeable to the user, outside of initial load times for latency exceeding a few hundred ms (up to a second or so delay if establishing a fresh TCP connection, depending on the latency).
Maybe TikTok Live could be an issue, but its a one-way channel so a delay shouldn’t be noticeable like it would be with a 2-way teleconference. Maybe if there was high jitter or packet loss, the stream could have extremely low quality.
- Comment on ISPs can charge extra for fast gaming under FCC’s Internet rules, critics say 7 months ago:
I would guess that since it mentions teleconferencing and gaming, that they want to create low-latency fast-lanes.
But they also mention TikTok, so I am not completely sure that they are referring to latency.
This article isn’t very good at explaining what they’re talking about on a technical level.
- Comment on Banana Pi BPI-M7 - More Reasons to Avoid the Raspberry Pi 7 months ago:
There’s an AliExpress link in the article that clearly prices it at $260…
- Comment on Apple argues in favor of selling Macs with only 8GB of RAM 7 months ago:
8 megs of RAM? I didn’t know they brought back the Macintosh II.
- Comment on Why AI is going to be a shitshow. 7 months ago:
- Comment on The Google One VPN service is heading to the Google graveyard - The Verge 7 months ago:
The only real use for Google’s VPN was as an extra layer of security when connected to an untrusted network or open WiFi.
- Comment on X’s Premium users can no longer hide their blue checks 7 months ago:
Nobody is unaware of that. It’s just that before Musk took over, the checkmark meant something completely different. What it means today is potentially embarrassing enough for people to want to hide it. Now they can’t.
Before Musk the checkmark meant “This user’s identity has been verified and they are notable enough to need that verification”.
Today it means “This user pays for Twitter”.
- Comment on Fairbuds are Fairphone’s proof that we really could make better tiny gadgets 7 months ago:
I can’t change my phone while using the adapter
Why should I need an adapter for something that used to come standard?
The same reason you needed a new charging cable every time the USB standard changes: because technology standards change over time.
If anything, audio adapters have been a thing for decades longer than cell phones have existed. This is not a new development at all.
- Comment on Movie industry demands US law requiring ISPs to block piracy websites 7 months ago:
Piracy websites should add a copy of the U.S. Constitution to their websites. Just slap a “/constitution.html” on the site.
Then, if the MPAA succeeds, we can talk about how the U.S. Government is blocking access to hundreds/thousands of copies of the Constitution online.
- Comment on Fairbuds are Fairphone’s proof that we really could make better tiny gadgets 7 months ago:
Most Bluetooth headphones are encrypted with a key shared only by the headphones and the host device. Not sure why you think they aren’t private. Maybe really cheap or really old headphones might not be so secure, but the vast majority of Bluetooth headphones in use today absolutely are.
Charging and audio quality are legitimate concerns, but again, you still have a headphone port… It’s just part of the USB-C port on the bottom of your phone. A $5 adapter completely absolves you of having to use Bluetooth.
I really don’t see how this is a big enough deal to care about.
- Comment on Fairbuds are Fairphone’s proof that we really could make better tiny gadgets 7 months ago:
I understand people’s desire for a headphone port, but I really don’t understand the hate for devices that don’t have one. It would be one thing if they vendor-locked Bluetooth headphones so you had to use theirs, but it really just seems like a common sense move in a world where Bluetooth reigns supreme.
As long as a USB-C adapter still provides the same functionality, I really don’t see anyone’s choices being taken away. If it is one less physical port on the device that helps streamline the hardware, I’m all for it too.
And if it is a dealbreaker, you don’t have to buy a fairphone.
- Comment on More and more people are ditching carrier roaming in favor of travel eSIMs 7 months ago:
Kinda nice that Google Fi gives you global roaming at no extra charge. Too bad it hardly ever works and text messaging is a shitshow.
Still used a travel esim on my last trip just to be able to reliably use my phone.
Probably dropping them soon because text message reliability is already a joke at home with them…
