General_Effort @ General_Effort @lemmy.world

Posts

164

Comments

1,906

Joined

2 yr. ago

Frauen werden deutlich seltener Opfer von Gewaltkriminalität: https://de.statista.com/statistik/daten/studie/1561134/umfrage/opfer-von-gewaltkriminalitaet-in-deutschland-nach-geschlecht/

Ein guter Teil dieser Gewalttaten dürfte wohl zu Hause erfolgen.

What's AmeBlog?

I had a look at what tumblr says and it's probably a good option. It's not likely that they will try to find sneaky ways around the settings. The liability risk is in no relation to the potential gain from selling that data. Under EU law, such opt-outs must be respected when training AI. For now, the major US companies can be expected to abide by that. In the future, we may see special models for the EU. A few open source models by Chinese companies already exclude use in the EU.

Reducing scraping takes skill and a major effort, which tumblr can bring. A catch is that there is a conflict between serving images to lots of people but not to scrapers. Sufficiently determined large scale scraping operations will still succeed, but maybe no one will feel that it's worth the effort anymore. It's impossible to prevent individuals from saving images. So AI hobbyists or small artists could still use your images for training and share the products. When fans re-upload your images, they may become part of large scale datasets after all.

Unhidden variables theory? Oh, please ...

It depends on what exactly you want to achieve. If you want money, then upload it to Adobe, Shutterstock, and such. That's the best offer you can expect. No one will offer you big bucks for an AI license just because you have protected your drawings.

If you want to have your own site, then you could rely on Cloudflare for handling the technical and legal side of preventing scraping/AI use.

But I guess the main worry for any artist is other artists who use AI. That's where Glaze and Nightshade come in. It's already been suggested but you should know how much you can expect.

These tools target the original Stable Diffusion 1.5. IIRC they also work on SD 2.0 because they reused some components. I am not sure what other versions, if any, could be affected. Certainly not the newer ones.

It goes without saying that the major companies were never affected, could not be affected. Since no one mentions it, I guess it's self-evident but I want to repeat it for the uninitiated.

I think these early models are still used partly because they have lower hardware demands and partly because they are less professionally censored (ie more suitable for porn).

Anyway, the effectiveness against hobbyists, your competitors, or other small scale AI users, is also limited. They may not use a susceptible model, especially if they make SFW images. If their model is susceptible, then these tools may waste a few hours of their time and maybe a bit of money. But it won't get rid of the competition or even significantly harm them.

Well, it looks like a jaw harp.

Hmm. Looks like a jaw harp.

I know in which hole the alcohol goes. I also know in which hole the young men go, at least according to "missionaries". Which leaves one hole for god, which also happens to be the one that people recommend when Jehovah's Witnesses bother them.

As a man of the world, I know, of course, that the god hole can also be fruitfully used for/by alcohol and young men, implying that Kayleigh was quite fun at parties. "Outnumbered", indeed.

If you've ever felt like some overly religious person was acting like they have a stick up their ass. Well, now you know.

Maybe it's time to have a little think about that word, "influencer", and how that is a job with which people make money.

Translation:

For when you have to move fast

The Böcker Agilo transports your up to 400 kg heavy treasures at 42 m/min - whisper quiet thanks to a 230 V electric motor.

It's also funny how Lemmy is buying up this narrative.

The entire US economy is currently being propped up by growth in the AI/tech sector.

What's happening is that Dementia Don is curb-stomping the US economy. AI investments, mainly in data centers, are the only thing that still seems promising. When you are on a trek and someone leads you through Death Valley, while pouring out all the water, you shouldn't blame the last horse that still keeps going.

Putting the blame in the right place would certainly help, with a view toward the mid-terms.

Financially: Diversify. Make sure that you are not completely dependent on what happens in the US. But mind that Europe comes with its own imponderable risks (ie Putin). Same with China. Maybe some old leader dies and the new crew runs everything into the ground; they go to war with Taiwan, that sort of thing.

GPT-4o mini via DuckDuckGo.

"so far"

Mit Sicherheit. Aber KI ist ein weiter Begriff. Da fällt mehr drunter als nur neuronale Netze.

Outage is ending.

Oct 20 2:27 AM PDT We are seeing significant signs of recovery. Most requests should now be succeeding. We continue to work through a backlog of queued requests. We will continue to provide additional information.

https://health.aws.amazon.com/health/status

Could be that your region relies on different servers, could be that only the sign-in is down.

I don't see what makes you so certain. The EU unambiguously wants computing devices to be more locked down. It wants responsible developers to be tracked.

If your argument holds, then that only means that there is a loophole allowing devs to distribute apps anonymously. That's where the car analogy fails. There are exceptions for small enterprises and "open source stewards". These exist so that small players and start-ups won't be overwhelmed by bureaucracy. They are not supposed to protect dev privacy or user freedom.

I can only repeat that I find your argument valid. I just don't believe it would stand up in court. If Google was pushing back on this, I would still back them up on such arguments. But they understandably don't.

Unless there is a major change in attitudes in Europe, we are going to see much more mandated control and surveillance, anyway.

If it’s bundled with the OS, it probably does.

Components "placed on the market separately" are explicitly included a being part of the product.

Let me try to gather this together:

The manufacturer, the authorised representative, the importer, the distributor, or other natural or legal person shall, on request, provide the market surveillance authorities with the name and address of any economic operator who has supplied them with a software product, including software or hardware components being placed on the market separately;

Economic operators shall, on request, provide the market surveillance authorities with the following information: (a) the name and address of any economic operator who has supplied them with a product with digital elements;

‘economic operator’ means the manufacturer, the authorised representative, the importer, the distributor, or other natural or legal person who is subject to obligations in relation to the manufacture of products with digital elements or to the making available of products with digital elements on the market in accordance with this Regulation;

‘product with digital elements’ means a software or hardware product and its remote data processing solutions, including software or hardware components being placed on the market separately;

The verification demand is for Google certified Android.

‘electronic information system’ means a system, including electrical or electronic equipment, capable of processing, storing or transmitting digital data;

The OS or a phone both fit that definition.

‘component’ means software or hardware intended for integration into an electronic information system;

An app fits the definition of a component.

Maybe you would have to argue that an app is not actually a component. But if it's a stand-alone thing, then why does it rely on an OS?

I think you can make a good argument that a phone without an OS is not a system. It's not capable of much. Maybe custom roms will remain an option.

Anyway, Google is not abusing that loophole. So, no problem. F-Droid encourages users to complain to EU lawmakers about Google being a meanie. Maybe the EU will close it anyway as part of future tech regulation.

Ein Warnhinweis:

Das Deepfake-Thema ist bekannt, oder? Viele befürworten Verbote. Schon mal überlegt, wie man die durchsetzt? Ohne Gesichtserkennung wird's schwierig.

Auch um "Rachepornos" auszumerzen, müsste man mit Gesichtserkennung, mit biometrischen Datenbanken, im großen Stil arbeiten. Allgemein: Um unautorisierte Fotos und Videos zu unterdrücken.

Vor ein paar Monaten ging eine Präzedenzklage gegen Google groß durch die Medien (unterstützt von Hateaid). In dem Fall geht es speziell um private, geleakte Amateurpornos. Es wird von "Bild-basierter Gewalt" geredet, aber das Ergebnis wäre natürlich auf alle Bilder/Videos mit Personenbezug anwendbar. Mit Gewalt ist die Verletzung des Rechts am eigenen Bild gemeint.

Google entfernt natürlich Suchergebnisse zur eigenen Person, die einem nicht passen ("Recht auf Vergessenwerden"). Man braucht keinen Grund dafür, um etwas entfernen zu lassen. Umgekehrt, Google muss begründen, warum es Suchergebnisse anzeigen oder auch nur speichern darf, per DSGVO.

Bei der Klage geht es darum, dass Google nicht nur gemeldet Links unterdrückt, sondern überhaupt alle "kerngleichen" Inhalte. So ein Wunsch lässt sich kaum erfüllen, ohne über das Aussehen der Person zu gehen. Andere Klagen zu solchen Überwachungspflichten sind da schon viel weiter.

Techfirmen, wie Google oder Meta, wollen solche Überwachungspflichten nicht. Sowas ist teuer und rechtlich riskant.

Wenn ihr also hört, dass "die Zivilgesellschaft" gegen die Techkonzerne klagt, um sie zur Einhaltung der DSGVO zu zwingen, denkt nicht, dass es gegen Überwachung geht. Um Daten zu schützen, muss man Daten überwachen.

Ich mach doch mal einen Factcheck.

"Gesichtserkennung" ist rechtlich verschiedentlich eingeschränkt, also nicht verboten. Im Umkehrschluss: Sie ist erlaubt.

Diese "Studie" bespricht, wie Gesichtserkennung im Internet funktioniert (zB bei PimEyes). Das scheint technisch solide zu sein. Aber die Behauptung ergibt sich daraus eben nicht.

Ganz kurz: Eine spezielle KI (kein Chatbot) analysiert Bilder/Videos und speichert das Aussehen der abgebildeten Personen als Zahlenkombi. Um effizient Personen auf verschiedenen Bildern abzugleichen, auf Bildern zu suchen, speichert man diese Zahlenkombis um arbeitet mit denen.

Im KI Gesetz heißt es (Hervorhebung von mir):

e) das Inverkehrbringen, die Inbetriebnahme für diesen spezifischen Zweck oder die Verwendung von KI-Systemen, die Datenbanken zur Gesichtserkennung durch das ungezielte Auslesen von Gesichtsbildern aus dem Internet oder von Überwachungsaufnahmen erstellen oder erweitern;

Wenn ich also eine Datenbank mit gesuchten Personen (vermisst, verdächtig, ...) erstelle, dann ist das ganz sicher unproblematisch. Wenn man diese Datenbank nun "Live" mit Überwachungsaufnahmen abgleicht, dann ist das kein Problem, solange nicht andere Vorschriften greifen. Problematisch wäre es, wenn die Daten über das Aussehen, die Zahlenkombis, von den zufällig erfassten Personen gespeichert werden. Dann hätte man eine Datenbank, aber wenn die Daten verworfen werden, dann ich sehe ich kein Problem in dieser Hinsicht.

Auch: Falls ausgesuchtes Material für spezielle Ermittlungen indexiert wird, dann sollte das eigentlich gehen, weil nicht "ungezielt".

Was genau "ungezielt" heißt, werden sicher die Gerichte klären.