Cryptik, to random

Now this is interesting. U of Chicago professors Ben Zhao and Heather Zheng have written an image distortion tool “Glaze” that actively poisons the training data for AI when artists’ work is scraped without compensation to train AI. And they’ve open-sourced it: Computer Science for Good (and a little mischief) https://www.technologyreview.com/2023/10/23/1082189/data-poisoning-artists-fight-generative-ai/

kkarhan,

@Cryptik I just doubt these tools can work simply by virtue of scrapers automating modern browsers and visually-impaired isers relying on CLI/TUI - based tools and screenreaders all too often.

It's like with : that gets solved faster and with less errors by than !

kkarhan, to random
pallenberg, to random
@pallenberg@mastodon.social avatar

Willkommen in einer Chinesischen Click-Farm, & damit dem besten Beweis fuer die These, dass der Like Button das Internet zerstoert hat!

Im November habe ich diese hier 👉 https://t.ly/LikeButtonLI ausfuehrlichst beackert... nur wurde dies vor den neuesten Entwicklungen in der Generative AI-Branche veroeffentlicht. Zeit also fuer ein Update:

👉 Social Selling Content-Farmen in Indonesien
👉 Streamer:innen unter Bruecken in Shanghai
👉 Filter, die den Druck der Schoenheitsideale verstaerken

kkarhan,

@pallenberg ist uralt...

Genauso wie ist das nix neues...

Und ja, es gibt neben Vollautomatisierung sogar extra mit Personen besetzte Clickfarmen um z.B. zu lösen und |ting Maßnahmen zu umgehen...

selbst spielt da ganz groß mit und vermarktet dies per !

mho, to internet German

Ist irgendwas im verhasster als ? Müssen ja was bringen, denkt man sich, sonst wären wir sie doch längst los. Denkste. lösen sie schneller und liegen öfter richtig:

Fast immer schneller, immer korrekter: Bots schlagen Menschen bei Captchas

Eigentlich wollte eine Forschungsgruppe ermitteln, wie lange Menschen für Captchas brauchen. Herausgekommen ist, dass Bots damit besser klarkommen.

https://www.heise.de/news/Fast-immer-schneller-immer-korrekter-Bots-schlagen-Menschen-bei-Captchas-9240739.html?wt_mc=sm.red.ho.mastodon.mastodon.md_beitraege.md_beitraege

di0v0n,
@di0v0n@graz.social avatar

@mho
Eigentlich war das Absichern der Logins mit nur eine zusätzlicher Nutzen. Im Grunde dienen sie zum Sammeln von Trainingsets für AI/KI bzw. Algorithmen.

Daher hat für mich die Studie gezeigt, dass die bisher gesammelten Trainingsets, zur Verbesserung der automatischen Lösungen geführt hat.

@heiseonline

jhpot, to random
@jhpot@mastodon.social avatar

This week I'm researching CAPTCHAs. Do any of you know anyone who works in that space? I'd love to talk to someone who is trying to keep these things working in the age of AI.

kkarhan,

@jhpot sadly noone I'm allowed to name, but never worked and are a blight as they are and only work at deterring users.

pluralistic, (edited ) to random
@pluralistic@mamot.fr avatar

Forget F1: the only car race that matters now is the race to turn your car into a digital extraction machine, a high-speed inkjet printer on wheels, stealing your private data as it picks your pocket. Your car's digital infrastructure is a costly, dangerous nightmare - but for automakers in pursuit of postcapitalist utopia, it's a dream they can't give up on.

1/

pluralistic, (edited )
@pluralistic@mamot.fr avatar

Today, there's a cool remote attestation technology called "" that replaces by having you prove to your own device that you are a human. When a server wants to make sure you're a person, it sends a random number to your device, which signs that number along with its promise that it is acting on behalf of a human being, and sends it back. CAPTCHAs are all kinds of bad - bad for accessibility and privacy - and this is really great.

49/

ciferecaNinjo, to technology in dsfawe

I print a lot. As far as I’m concerned, is dead. It was killed by MS & Google. Probably 99+% of the world still does not use PGP. Web forms? No, they’re also dead to me because I refuse to solve most & most certainly will not solve a Google . Message centers? Also dead to me because a good number of them are snooped on by Cloudflare, or they proactively block Tor.

Even if a web form or message center is non-Cloudflare & open to Tor users, most of them demand too much info. They always make email address a required field. They’re not getting an email address from me if their MX server is Outlook or Gmail.

Hence why I’ve gone back to the paper letter, apart from the few recipients who still have a fax number.

cs, to random
@cs@mastodon.sdf.org avatar

I, a human being, have not been able to get by the that are sprinkled on this web page. I think they actually originate from the embedded YouTube videos, but I really can’t tell. 🤷‍♂️

ppatel, to fediverse
@ppatel@mstdn.social avatar

People of the , especially those creating various services, there are alternatives to HCAPTCHA when you ask people to create accounts or login. Despite HCAPTCHA's propaganda, HCAPTCHA is not truly accessible to people. It requires us to give up our privacy if we choose the cookie option. The text alternative doesn't work most times. I just tried to sign up for a /kbin server at https://redit.buzz and wasted 30 mins on HCAPTCHA.

Please be !

kkarhan,

@ppatel in general should be outlawed as features!

AFAIK, @stux combats via verification eMail and rate-limiting registrations...

hen, to random
@hen@social.lol avatar

This morning I was pretty aggressively captchad on Brave Search. Literally couldn’t get past it either. Recorded a video and posted on our forum to see if this is at all common: https://discuss.techlore.tech/t/any-other-brave-search-users-getting-capthad-i-couldnt-even-search-this-morning-video/3888

Has anyone had this issue? #brave

kkarhan,

@lo__ @hen also #Captchas are actually #ableist and not only require #tecking in the form of #cookies to work, but #bots are even better than #humans, so just telling people to slow down will already work most of the time...

#DDoS'ing bots can be filtered behaviour-based...

jake4480, to ai
@jake4480@c.im avatar

Nobody asked what's worse than regular captchas, but here's the answer: having to try and solve ai-generated nonsense captchas. Find those Yokos!

maxleibman, to ChatGPT
@maxleibman@mastodon.social avatar

Breaking ChatGPT’s spirit by just asking it to solve captchas over and over again.

i0null, to random

:itsaliens:​

  • All
  • Subscribed
  • Moderated
  • Favorites
  • megavids
  • mdbf
  • ngwrru68w68
  • modclub
  • magazineikmin
  • thenastyranch
  • rosin
  • khanakhh
  • InstantRegret
  • Youngstown
  • slotface
  • Durango
  • kavyap
  • DreamBathrooms
  • JUstTest
  • normalnudes
  • osvaldo12
  • tester
  • GTA5RPClips
  • cubers
  • everett
  • tacticalgear
  • ethstaker
  • provamag3
  • anitta
  • Leos
  • cisconetworking
  • lostlight
  • All magazines