Show newer
Olives boosted

When someone complains about porn one minute, complains about end-to-end encryption the next, and advocates censorship the next, that doesn't breed confidence in me that their argument that human rights concerns with undermining end-to-end encryption could be fixed with the so-called mythical government which never over-steps it's boundaries has much merit.

Olives boosted

theverge.com/2024/2/22/2408013 Avast fined for selling customer info without their consent and lying about it.

sahanjournal.com/business-work Amazon coffee machine under fire for secretly taking pictures of employees in breakroom.

reclaimthenet.org/eu-group-loo E.U. working group apparently looking for ways around ECHR ruling prohibiting undermining end-to-end encryption.

edri.org/our-work/press-releas EDRi pushes for ban on spyware after politicians were attacked by phone hacking malware.

reason.com/2024/02/22/proposit San Francisco police to be able to operate any method of surveillance for a year before review. San Franciscans have a chance to vote against it.

reclaimthenet.org/biometric-en Major League Baseball is scanning your face.

reclaimthenet.org/maine-school School drops plans to fingerprint students.

reason.com/2024/02/20/nyc-chil Child Protective Agency coercing parents into allowing them to search their homes without a warrant.

reclaimthenet.org/facial-recog New Zealand's Privacy Commissioner helps grocery store to invade citizens privacy with new face recognition program.

@itwasntme223 I'm fairly sure that the "Nazis" are largely right wing leaning trolls who do that to be provocative.

It is kind of ironic, because on a few occasions, they seem like the sort of people the Nazis would have taken away to a concentration camp somewhere (obviously not good but... should have read "first they came for....").

That said, there seems to be a bit of an obsession with them. Sometimes, chasing shadows. There is also a certain amount of admin toxicity and almost toxic narcissistic type behavior there.

Olives boosted

I take solace in the continued evidence that generative #AI models invariably produce obvious artifacts in their output. I think that may be just the nature of neural networks; they're probabilistic, and the chance that their output doesn't include something weird is very small.

Just take a look at every #sora video the AI-bros are freaking out about today, and you'll see each one has at least one obvious tell, if not multiple.

Olives boosted

Slapping the #AI label on generative models is a move designed to make them seem mysterious and powerful, and thus attract media hype and venture capital. We can see this an intentional weaponization of that Arthur C Clarke quote about advanced technology being indistinguishable from magic.

Evil vending machine scandal making the rounds, lol. It is troublesome though that it has a camera hidden inside it, and that it's recording people.

theguardian.com/world/2024/feb

"A malfunctioning vending machine at a Canadian university has inadvertently revealed that a number of them have been using facial recognition technology in secret.

Earlier this month, a snack dispenser at the University of Waterloo showed an error message – Invenda.Vending.FacialRecognition.App.exe – on the screen.

There was no prior indication that the machine was using the technology, nor that a camera was monitoring student movement and purchases. Users were not asked for permission for their faces to be scanned or analysed."

@devnull I've heard of that happening to instances. They'll wake up one day and they'll have a nasty surprise $10k bill from something stupid.

@devnull If you are relatively liberal about what comes in, it's probably less of an issue, but boosts do increase the moderative surface area a bit (and trying to moderate it can lead to absurd looking situations).

@devnull I suspect it's because if something goes "viral", then there might be a stampede of requests to that resource, which might accidentally wind up DoSing an instance.

Actually, though, there are quite a few instances (or there was) which don't host resources themselves, and which load them from the original instance, although they seem to be on the smaller side.

From a liability perspective (although, there is already a certain amount of lee-way there), linking is superior to proxying content.

@thenexusofprivacy The fediverse has bigger issues with admin toxicity, people being parachuted in, (and to a lesser degree, QAnon) right now tbh. This is a social issue which cannot be straightforwardly fixed with technology.

One of a number of reasons why I'm not motivated to put development time into the fediverse.

That said, if what someone wants is a high level of safety, as it were, I don't think federation (or not the model of a community glued to a server) is an answer to that. I came to that conclusion after watching people trying to "fix" the fediverse for the past year.

en.wikipedia.org/wiki/KISS_pri Seeing someone coming up with grandiose (and probably harmful) fedi bureaucracies in their imagination reminds me a lot of this.

The anti end-to-end encryption suit which tries to lean on "deceptive practices" just shows that the government will try to lean, hard, on any tool they get, which is just another sign of why KOSA wouldn't be good.

@riana I think the realism argument is typically exaggerated (even the people who argue it tend to contradict themselves). Even in that case, I'd find a "War on Drugs" type tending to violate civil rights prohibition approach to be suspicious. It's not clear it wouldn't end up doing more harm than any possible good.

I think the most effective thing might be to deal with conduct points (i.e. sextortion, harassment, maybe invasion of privacy), as these seem to be what people are worried about, rather than making broad content based prohibitions (which tend to violate human rights, and likely involves "non-existent people"). It's not really effective, proportionate, or sustainable, and it would drive the surveillance / carceral state, and the harms inherent to that.

Also, the contexts in which someone has cited it being present (i.e. criminal sites which already clearly violate 2251) would suggest they could already be prosecuted for that other criminal conduct.

It is also not clear it is a significant issue. It seems exaggerated, and it seems every month, someone is arguing the end might be right here, although with hardly much new to argue with. Examples are usually like "this site or that site", and often, there is language which quickly makes it sound a lot like "non-existent people".

That said, I don't necessarily disagree with your suggestion for authenticating the image provenance. And perhaps, if something is less "black market", then there might be more incentive for a service provider (which deals in such things, not a mainstream one) to assist in that, than if they're liable either way.

Also, in that other country where it seems to be protected speech, the arguments look a lot the same. It is argued that a prohibition would violate the Japanese Constitution which protects free expression. As far as I know, the Supreme Court there has ruled that it could be restricted in the case of something depicting actual children.

Instead of making people prove whether they're human, maybe we should make them prove they are machines, lol.

Show older
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.