We're partnering with Braag and Hyperstarter to re-launch our campaign with even better outreach. This campaign isn't going away, it's leveling up. Please see the Updates page for details!
We'll soon be posting a signup for our waitlist at waitlist.havensocial.org - please join the waitlist so we can keep you in the loop and make sure you get first access when we're back.
The waitlist page will fill you in about our free art commission raffles, so don't miss out on those!
To our wonderful backers: As a thank you for your support, we're giving you 25% off everything in our next campaign. Check your Kickstarter messages for more information about this. If you don't have a message from us, but think you should, shoot us a message on Kickstarter and we'll get it figured out.
It's a bleak time to be on the Internet.
Social media megacorporations have entirely enshittified themselves, and work to exploit, manipulate, and spy on their users.
Oligarchs and governments modify the algorithm of social medias they own, to manipulate elections and public opinion towards what benefits them, but harms you.
Creators on these sites are subject to the whims of a fickle, inscrutable algorithm, and have to spend hours engaged in useless platform Kremlinology to avoid getting shadowbanned.
Floods of AI bots steal any written work or art they can find in order to mass produce cheap imitations, and send any pictures of your face they can find to Big Brother-esque facial recognition mass surveillance dragnets.
Haven's here to protect you from that.
We're building an anti-AI platform that takes YouTube's support for long-form landscape video and playlists, Instagram's support for images and reposting, and combines them in a UI that seamlessly shows both types of content. You'll be able to easily swap between a discovery algorithm, and a reverse chronological following-only algorithm. In the video above, I say TikTok, but because of the tremendous amount of community feedback we've decided to focus on replacing YouTube and Instagram instead. Our platform will:
Protect users from invasive surveillance firms like PimEyes and Clearview.AI, the company that famously scraped billions of pictures off of Facebook to use in a facial recognition surveillance dragnet. We do this by modifying all images of users in a way that makes it indecipherable to facial recognition models, while not appearing any different to the human eye.
Protect artists from having their art stolen by generative AI art models without copyright. Every image uploaded to our platform is automatically "poisoned" against AI, with no effort required from you. If an AI ingests the poisoned art to use as training material, the AI cannot learn anything from it, or worse, learns wrong. This gives artists the power to enforce their creative rights! If AI companies pay the artist for the unpoisoned version, they'll be fine. But if they use the art without the artist's permission, the AI model is sabotaged. We’re essentially setting landmines for AI. Artists are a majorly influential faction online, and we’ll be the best place for them to post.
Poisoning art is out of reach for many because it requires an expensive graphics card and technical know-how – Haven provides those for you, making it easy and free for anyone to protect their art and biometrics.
Right now, we're using Mist to poison images, and planning to make regular donations to the Mist team. We’re currently pursuing permission to use Glaze and Nightshade as well. (but to be clear, we don't have it at the moment. For now, you should evaluate this project as if we will only use Mist.) Stay tuned for updates!
Free users from the severe censorship and algospeak enforced by today's platforms. Needing to use words like “unalive”, “music festival”, "le dollar bean", and "Ta!wan" instead of what you actually want to say is a idea pulled straight out of dystopian fiction. We're creating a place where people don’t have to worry about being shadowbanned for talking about topics important to them.
To be clear, even though we are anti-censorship, we are not free speech absolutists. If you pursue absolute free speech, your platform turns into 4chan, and we don’t want that. We mean anti-censorship more like protecting people from:
Shadowy megacorps and governments promoting or suppressing certain news stories to suit their goals.
Oligarchs modifying the algorithm of social medias they own, to manipulate and propagandize you into having worldviews that benefit them, but harm you.
Easily move your content and audience across platforms. The annoyance of downloading all your art and re-uploading it somewhere else makes many people hesitant to switch platforms. But with Haven, it couldn't be easier. If you ask us to, we’ll go retrieve all your photos/videos from Instagram, YouTube, Twitter, etc, and automatically upload them to your Haven account, and backdate each post to match the date of when it was uploaded to the other site.
We also make it so your followers don’t have to re-find you on a new site. When creators automatically move their posts over in bulk, we’ll keep internal records of which Instagram/YouTube/Twitter account corresponds to which Haven account. So when a user signs up for Haven, if they ask, we can go look at their Instagram/YouTube/Twitter following list, translate those to the corresponding Haven accounts, and automatically make them follow each creator on the list. So, nerd speak aside, we essentially move your following list from one site to another.
Compensate our creators better than anywhere else. We’ll automatically detect and remove as much AI-generated content as possible, but AI art detectors are unreliable in general. That’s why we have another trick up our sleeve to amp up our manual detection of AI content, and pay our creators more at the same time: crowdsource the detection and turn it into tips.
Have you ever been on Snapchat or TikTok when it shows you a little survey asking about the content you just watched? We provide rewards for users when they answer those for us. So as you scroll, we’ll occasionally ask: “Do you think the previous post was AI slop? Does it break our site rules? Was it adult content, but wasn’t marked as adult content?” Scroll past it if you'd like, or answer it and we’ll give you a dime, quarter, or similar, for you to donate to a creator that you enjoy. This is not a position or job to be held, like a Reddit mod. Rather, we ask a random sampling of viewers each time. It’s not a lot of money, but you’re also only spending 5 seconds to get it, so it aggregates quickly. If a creator has, say, 50k subscribers, each giving them one dime a month, that’s $5,000 a month, on top of what the platform itself pays you.
This whole scheme allows us to reduce how much work our content moderation staff have to do, and we can give the money we saved on salaries directly to our creators. It frees up a good chunk of our budget, plus it fosters a sense of community.
We're following the Steam playbook of "total market domination by being nice and treating users better than your competition" with practices like:
No Monetization Abuse On YouTube, if one video goes viral and it alone carries you over the monetization threshold, you get nothing, because you’re not monetized yet. So then you have to go viral all over again! :( With Haven, once you hit that threshold, Haven remembers at what point you hit it. So once you apply for monetization, we’ll go back and say “Yeah, your video pushed you past the monetization threshold when it only had 1/4 of the views it does now. So, we’ll backpay you for that other 3/4 as if you were monetized.”
No Copyright Abuse On YouTube, a copyright claim immediately sends 100% of revenue to the holder, because YouTube doesn’t care enough to spend the time to see if it could be Fair Use or stick up for you. They immediately cave to the claim.
Haven will be on your side and try our hardest to classify your videos under Fair Use, or make a deal with the copyright holder so you only pay a few percent of the video’s earnings to the holder. Losing a few percent is a bummer, sure, but it’s better than losing 100%!
Protect users from manipulation by providing a transparent recommendation algorithm. Do you want the beloved reverse chronological feed back? You’ll get it with us!
Most platforms force their creators to be at the mercy of a fickle, inscrutable, manipulative algorithm. You may already know from experience how this adds a lot more stress and work! It’s frustrating to make good videos and have the algorithm basically shadowban them because you tripped an invisible tripwire! We make our algorithm's inner workings very clear and well-explained, to help creators develop a trusting relationship with us.
Promote healthy use of social media. Social media companies want you to doomscroll, and they want you to be angry! It makes them more money! But I believe this is dangerous and self-destructive for society. They’re making people constantly tense and scared and worsening everyone’s mental health to make cash. To prevent this, we're building the best time limit functionality into Haven. Check the thesis for more on this.
For the same reason, our algorithm will also promote quality content over engaging-by-any-means content. On other platforms, creators will intentionally post enraging, drama-stirring, exaggerated-strawman content because any interaction, even negative interaction, counts as engagement, and more engagement means more algorithmic boost. Haven prevents this by promoting posts based on the amount of positive engagement with the post, not total engagement. Specifically, we avoid feeding the trolls by:
not counting comments as engagement, since that's the main place people voice disagreement or confusion
making it so disliking a video/image revokes any algorithmic boost it gained from your engagement. It's as if you never watched the video. So you get the last laugh against clickbait and ragebait.
Provide better DMs and comments. Social media direct messaging has often been riddled by the same issues across different platforms, which is a shame, since these issues aren't all that difficult to fix. At least, not compared to everything we've described previously. Note: these features will likely not be added until we add DMs in a future update. We're being cautious about feature creep, and focusing on the core features of Haven first to get it to a point of public release.
Block Images: On Haven, if you'd like, you'll be able to control who can DM you images and who has to stick to plain text. For example, maybe only people you follow are by default allowed to send images, or maybe every new person DMing you can't until you specifically enable it for that chat. Or maybe everyone can! Ride the lightning! Up to you. This helps you avoid seeing... unsavory pictures... by surprise.
Sentiment Filter: Other platforms filter by who is able to message you, which is helpful, but can only work in broad strokes. On Haven, you’ll be able to do this, as well as filter who can comment or DM you by the content of their message. If you enable it, (and only then), we can take all new incoming DMs, run them through sentiment analysis, and only allow the normal or nice DMs through. It's like a digital secretary that throws away the hate mail and death threats before they reach your desk. To be clear, we don't expect this to stop all mean DMs, but it's better than nothing. Additionally, you don't have to turn it on - your inbox can be the Wild West if you wish. See the FAQ about this as well.
We're making an educated guess that doing this will be incompatible with end2end encryption, but not to worry, we're planning for that. If you have sentiment analysis turned off, end2end encryption is enabled from the very start. If you have sentiment analysis turned on, then it turns off and end2end encryption turns on as soon as you accept the DM request.
How Will Haven Fund Itself Long-Term?
Great question! We answer this in our interview with VildrArt & inkiio and Roxane Lapa. In brief: we run ads. But our ads are far more ethical, and serve the needs of the users far more, than ads on any other platform. Please check out the interviews to see what we mean by that!
Now, Haven adds a lot of extra features on top of the basic model of "image/video based social media", and each extra feature costs money to run, so you might be thinking: can ads truly possibly pay for all those costs? The answer is yes. Ads can be extremely lucrative - Facebook, Instagram, and YouTube are all basically entirely funded by ads, and still have the budget to spend incomprehensible sums of money on AI projects. For example, Facebook is planning to spend 600 billion dollars on AI over the next 3 years. This is to be expected; they love picking a new way to waste money and get nothing out of it every few years - remember when they spent 70 billion dollars on the Metaverse? They do something like this every few years, and Facebook is still massively profitable.
Similar money pits exist for each of the other platforms, showing that there's a pattern here: once a social media gets to a certain size, they get to pick a never-profitable pet project, like AI, the Metaverse, etc. Each platform has their 'thing'. Providing computing power for art poisoning and funding R&D of new poisoning techniquescan be our thing. Since we won’t be wasting revenue on AI hype, that frees up some budget to spend on other costs, like this :)
We're similar to Steam in this regard - Steam's 'thing' is providing an extremely lenient return policy. There's really no need for Steam to provide such generous return policies - without them, Steam would still be leagues better than their competitors. But Steam provides them anyways, even though it means incurring extra costs. It's the same with us.
In the same way that Facebook is massively profitable despite incurring compute costs, by running a massive amount of ads... Haven can be reasonably profitable despite incurring compute costs, by running a reasonable amount of ads. (and yes, we know that ads mostly only work at scale - we are confident Haven can get to a scale where that's feasible. Especially so with your support!)
What Your Support Will Do
Help us grow our team with more experts.
Buy hardware. The software we use to poison your selfies and art against AI scrapers requires some beefy graphics cards - and a lot of 'em in order to do it at scale.
Fund ongoing development and improvements.
Every donation past our goal helps us finish development sooner, and get more hardware to protect more images, faster.
Get Involved
You can learn more about enshittification and how social media is used to manipulate and spy on you by checking out our founder's Master's thesis here: https://docsend.com/v/pgw4q/haven-thesis
Please join our waitlist at waitlist.havensocial.org so we can keep you in the loop and make sure you get first access when we relaunch our campaign. The waitlist page will fill you in about our free art commission raffles, so make sure to grab your chance at one of those! Here's an example of one of the raffles:
If you have any questions, feel free to message us on Kickstarter. It's always fun to teach and share knowledge.
Do you know a YouTuber, journalist, artist, musician, actor, author, producer, or other public figure who would like to help us spread the word about Haven? Shoot us a message on Kickstarter or our website. We'd love to chat, sponsor you, do an interview, or otherwise work together!
The Ask
We’re going to protect people against the evil of Big Tech megacorporations and surveillance states, but we can't do it without your support. Could you chip in to protect your privacy, and protect artist's livelihoods?
Together, we'll build a platform that protects your face, your art, and your privacy from invasive AI dragnets, and provides creators a way to escape the abuse of Instagram and YouTube.
A Special Thanks To
We couldn't have done this without some help from the amazing YouTubers we've sponsored. Please do go patronize their channels:
In software development, it's often difficult to estimate how long it will take to finish a particular feature. We'll be sure to avoid the do-nothing meetings that plague most software teams, freeing up lots of time and speeding up development.
I'll also publish a dev blog to share recent successes and interesting happenings as they occur, to keep backers updated.
The challenges we've faced have prepared us to excel - for example, our previous failed Kickstarter. With our previous campaign, we were still figuring out the Kickstarter platform, and did not do nearly enough advertising.
This time, we're MUCH stronger with our outreach. 15+ YouTube sponsorships and counting, with channels in the millions of subscribers. We're on podcasts. We're at cybersecurity conventions. We're running raffles for free art from the YouTubers we've sponsored - check our socials for those! The difference is night and day, and we're incredibly excited.
Mass spread of poisoned art will reduce the quality of AI art generators, which will reduce the amount of people using them, which is GREAT for the environment :)
Here's a very interesting paper about how even small amounts of poisoned data can have a large impact, no matter the size of the AI model. It's published by no less than the Anthropic AI team themselves, who would have every incentive to downplay the results. It's important to note this paper is about text-based poison, not image-based poison, but we think it is still an interesting lead.
www.anthropic.com/research/small-samples-poison
I'll share with you my curated playlist of music that helps me focus while programming! 400+ energetic-but-not-overwhelming, instrumental-only songs and counting. The lack of vocals helps them fade into the background while you concentrate.
Small donations have a huge impact when we all lift together!
5 cybersecurity and privacy themed stickers made exclusively for this campaign + my playlist + credit on our site. The music sounds better with more stickers on your laptop.
Backers
73
Ships to
Anywhere in the world
Estimated delivery
3 items included
5× Stickers, Your Name in Credits, My Programming Jams Playlist
We plan to pay our staff by running (very few!!) ads. But you're above that. Your account will be ad-free for 3 years, plus everything from previous tiers!
Please see the FAQ to see how running ads allows us to fund development of new ways to protect your privacy and art.
Backers
49
Ships to
Anywhere in the world
Estimated delivery
4 items included
Ad-Free Account, 5× Stickers, Your Name in Credits, My Programming Jams Playlist
With this tier, you can pick a username right now, today, before anyone else gets to. Grab exactly the username you want before someone else does!
After you back this tier, please fill out this short form:
https://forms.gle/HcYLy4VwabpM3Tfs8
In the very small chance that two backers pick the same username, the person who backed and filled out the form earlier will receive it.
We believe it will be easier to manage and less stress to have backers pick their usernames as they trickle in, instead of sending out an email to everyone all at once after the campaign.
This tier includes all rewards from previous tiers.
Backers
61
Ships to
Anywhere in the world
Estimated delivery
5 items included
Reserved Username, Ad-Free Account, 5× Stickers, Your Name in Credits, My Programming Jams Playlist
This tier is for those whose are unrivaled in their desire for privacy and protection. As our deepest thanks, we'd like to give you every item from every tier, in addition to exclusive rewards: an ad-free account *forever*, and a free copy of every new sticker we ever make going forward. It's passive income, but stickers.
"Every item" includes both a forever and 3-year ad-free account, both with reserved usernames. Give one to a friend!
Feeling extra generous? Select this tier and for every additional $50 you donate, you'll get another ad-free forever account, with a reserved username, and another 5-pack of stickers. We would offer this as an add-on, but unfortunately we can't restrict add-ons to only certain tiers.
Please also look at the Reserved Username tier to see how to lock in your username!
Please see the FAQ to see how running ads allows us to fund development of new ways to protect your privacy and art.
Backers
244
Ships to
Anywhere in the world
Estimated delivery
7 items included
Every New Sticker - Forever!, Ad-Free Account - Forever!, 2× Reserved Username, Ad-Free Account+3 More
