Baby Elon Musk Shrimp Jesus. The
Titanic colliding with an iceberg lettuce. Social media is flooded with AI slop – low-quality, often absurd and sometimes disturbing images, videos, and words created by generative artificial intelligence.
Some AI slop may be obvious fakes, but a lot of it isn’t. I remember the very first time AI fooled me. I was struck by a video of Snowy owls in the Arctic with a brood. I immediately shared the image with my wife, who was equally amazed. Her response was “Surely that can’t be real?”
No, it wasn’t. Quick review of the poster Yournaturescenary, which has more than 230,000 followers on its platform, revealed that the video was in fact AI-generated. I felt deceived and suspicious of all images I saw online after that.
I used to use Instagram for following creators and sharing my own photographs of wildlife and nature — you know, those images that I captured with an actual camera and my human hands. Recently, scrolling down my feed has been more about spotting the bot than enjoying it. It’s annoying, yes, but AI images that look real can also cause severe harm.
The of EU tech
The latest rumblings on the EU tech scene
A story from our wise old founder Boris and some questionable AI artwork. Every week, it’s in your inbox for free. Sign up today!
This has forced me to find an alternative. It turns out I’m not the only one.
Jingna Zhang, founder and CEO of Cara, a social media app decentralised for artists, told TNW that the most common feedback she receives is that people feel overwhelmed by AI-generated pictures flooding mainstream platforms.
Unfortunately, the problem is likely to get worse before it improves. Meta opens the AI floodgates.
Meta, parent company of Facebook, Instagram, has embraced AI, scrapped fact-checkingand plans to allow bots to open their own accounts. Meta started labeling AI generated content last year. However, its algorithms have been shown to mistakenly label real images as bot-created images and vice versa. The platform also makes it easy to disablethe label.
But what’s the cherry on top? Meta uses publicly posted posts to train its algorithm. Meta opts all user accounts in by default, even thoughthe majority of European users are protected by data protection laws. AI algorithms are improving rapidly, fueled by all the free data —hurting artists and creators . It’s the same as learning valuable skills from your friend, using them to start a business together, then cutting them out. “19459088” ring any bells?Zuck?
AI slop and bot accounts, with little or no moderation, are all part of what Cory Doctorow called the ” “” enshittification “of the internet. While many creators have accepted the new reality, others are seeking refuge on safer corners of the Internet.
Cara is one of the winners. Meta’s app grew from 40,000 users to 650,000 in one week when it announced plans to train AI on user content.
Zhang says, “They [artists] were looking for a place to experience genuine human-tohuman connection in online communities and art. That’s how they discovered Cara.” Zhang says that it is essential to create new rules for online creators.
Zhang says, “We believe the ethical and data privacy concerns around these [AI] databases urgently need to resolved through regulations. Governments should clarify their positions or adopt bills to protect all individuals and artists from such unauthorised uses.”
This is a problem that Zhang, a Singaporean photographer who has won awards, is all too familiar with. She was successful in a landmark appeal before a Luxembourg court over a painter that copied one of her photos. She is also suing Google for allegedly using her photos to train its AI Imogen model without consent. She is also a plaintiff in similar cases against Midjourney Stability AI Runway AI and DeviantArt. Zhang launched Cara as a refuge for artists who want to connect and share in 2023. This volunteer-run organisation is available on both a mobile app and a website. Users can post images on their profile and engage with others by liking and commenting. They have a feed, where they can post their updates, similar to X or BlueSky. The platform has currently 1 million users.
Cara’s biggest selling point is its AI policies. The company told me that all users are automatically opted-out from AI bot scraping by default. Cara does not host generative AI on its platform and relies on a combination of human moderation and automated detection to keep it clean.
For Zhang preventing bots from being used on the platform is not just about piracy by users, but also protecting art. “What makes art meaningful to me is knowing that someone took time out of their life to practice, build skills and express themselves through art,” says Zhang.
How does it compare to Instagram? My first impressions of the app are positive. The app is easy to use, has a layout similar to other social media sites I’ve used and offers content of high quality. I can relax, take off my bot-spotting sunglasses, and enjoy the art, knowing it was created by humans.
Cara is clearly for artists – sketchers, painters and graphic designers – not photographers. Pixelfed was the app we used to solve this problem.
It’s not perfect, but it’s not Instagram
Pixelfed, a photo-sharing app similar to Instagram, is run by Canadian developer Daniel Supernault. It is open-source and decentralised, but without algorithms or ads. AI-generated content, like Cara, is prohibited.
Like Instagram, Pixelfed allows users to send direct messages, share photos and videos and discover new ones. The feed is chronological, so posts are displayed in order of publication. This is similar to Instagram before 2016, when the feed was based on an algorithm that showed you what it thought you wanted to see.
Pixelfed, despite only having 700,000 users, already hosts a lot of great photography. The experience is unfiltered and raw, and the AI-free and ad-free environment allows you to focus solely on the content.
Pixelfed is powered by ActivityPub. This open protocol was developed by Tim Berners Lee’s World Wide Web Consortium. ActivityPub is also used by other apps, such as Mastodon (founded by German developer Eugene Rochko ) and YouTube alternative PeerTube. Together, they form the fediversea decentralised system that allows users to interact, follow, and share across different apps – similar to how email works between Gmail and Outlook. The
Pixelfed user base grew quickly after the release of their app on iOS and Android. It was the most downloaded social app on Google Play for a short time after its launch in the US and France.
With Supernault as the sole developer and the surge in traffic there are bound to be some bugs. One of the bugs was that my Android phone’s back-button didn’t work. I had to reach up to the top left to use the button in the app. The app is also a little slow.
Supernault raised over $90,000. That might help him fix some bugs or even hire another developer or two.
But, as oneReddit users put it: “People move to Pixelfed not because its perfect but because it isn’t Instagram (I think anyone needs an explanation at this point).” If you don’t feel that staying on Instagram will affect your values,[,] then don’t move.
Goodbye, Instagram. Will you join me, dear friends?
