AI firms like PimEyes and Clearview.AI scrape billions of photos of people off of the Internet, then shovel them into AI facial recognition surveillance systems to track your every move. Haven protects you by modifying all footage of users in a way that makes it indecipherable to facial recognition models. All media you upload is safe from Big Brother's prying eyes.

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.

Every video and image uploaded to Haven is automatically "poisoned" against AI. If an AI ingests the 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 setting landmines for the AI.

Poisoning art is out of reach for many because it requires an expensive graphics card and technical know-how. But with Haven, you can use ours for free. We make it easy for everyone to protect their art, livelihood, and biometrics. 

Haven makes it possible for art to be poisoned at a massive scale. As poisoned art is shared, it will become densely spread across every site, and even better, it's unmarked. AI art scrapers will quickly learn to avoid Haven, but our hidden landmines are spread internet-wide. The scrapers will have no safe harbor.

Right now, we're using Mist to poison art, 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!

Social media companies want you to be angry. When you're angry, you doomscroll more, and make them more money. But we believe this is dangerous and self-destructive for society. Agitating users every time they open the app leaves everyone constantly tense and at each other's throats.

To prevent this, Haven's algorithm will promote quality content over engaging-by-any-means content. On other platforms, creators intentionally post irritating, deceptive, drama-stirring content because any interaction, even negative interaction, counts as engagement, and more engagement means more algorithmic boost. Haven prevents this by promoting videos based on the amount of positive engagement with the video, not total engagement. Don’t feed the trolls.

Do you want the beloved reverse chronological feed back? You’ll get it with Haven.  

Most platforms force their creators to be at the mercy of a fickle, inscrutable, and manipulative algorithm. Figuring out how to please the algorithm can take just as much effort as creating the rest of the video! It's frustrating to make quality videos and have the algorithm shadowban them because you crossed an invisible tripwire. We ensure our algorithm's inner workings are clear and well-explained, to help creators develop a trusting relationship with Haven.

.

Our creators will be compensated better than any other platform. In addition to the normal creator compensation, we give users rewards when they help us categorize posts or report rule-breaking posts, which they can then give directly to their favorite creators. 

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, 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/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/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. I explain this more here, at timestamp 42:10.  

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 do that. 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?” Answer that, and we’ll give you a dime or quarter or similar, for you to donate to a creator that you enjoy. It’s not a lot of money, but you’re also only spending 5 seconds to get it, and 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. I talk about this more in depth in this interview with VildrArt at 14:26.