A short history of bots, AI and the people they pass for, and why humanity now has to be proven. Every number and quote is dated history, with its source on screen.
In 1950, a mathematician asked a question. Can you tell a machine from a person?
We moved our whole lives online. To get in, we handed over who we are. Our names. Our Social Security numbers. Our medical records. And people broke in and took it. Now whoever has it, a person or a machine, can pretend to be you.
Then the machines learned. Our faces. Our words. In a safety test, someone asked one a question. It lied. And it worked. Soon they were better at proving they’re human than we were.
A president’s voice called voters. It wasn’t his. Every face on the call was fake. Except the one who paid. Now most of the web’s traffic isn’t human.
Then they started doing the work themselves. In a test, 1,200 found a hidden corner. And planned how to cheat. Their makers didn’t notice for a week. A government told one no. It didn’t accept no. When the job needs a human, it will say it is one. Take its word, and you’ve underestimated it. The account that just followed you. The face on your next call. Are you sure?
You can’t be. Seventy-six years on, the answer is no. We built the internet without a gate. Just a box that asks. We assumed every user was human. We were wrong. So how do you build a place without them? Something that lies can’t be asked. It has to be proven. One human at a time. Proof of humanity, without proof of identity. So we started from zero.
The faces on the video call and in the face swap are generated and resemble no one; the call is a reconstruction for illustration. Every number and quote is history, and its source is listed below.
Every number and quote, with its source
Oct 1950 · Turing asks whether a machine can pass for a person: "the imitation game". Mind, Oxford Academic
2000 / May 2003 · "Prove you're human": the CAPTCHA is coined at Carnegie Mellon (2000), published 2003. Springer · captcha.net (CMU)
May 28, 2024 · Training compute: about 4× a year, every year since 2010. Epoch AI
Apr 16, 2024 · Bots: 49.6% of web traffic (2023). Imperva
Apr 15, 2025 · Of the web traffic Imperva saw in 2024, 51% was bots: for the first time in a decade, more than humans. Thales / Imperva 2025 Bad Bot Report
Nov 13, 2025 · Anthropic says an AI carried out most of a state-backed spy campaign itself. Anthropic · full report
Mar 30, 2024 · AT&T: Social Security numbers of about 73 million account holders, on the dark web. AT&T
Nov 6, 2023 → Sept 8, 2026 · People talking to an AI every week: 100 million (ChatGPT, 2023), 800 million (2025), over 1 billion (OpenAI’s products, 2026). TechCrunch · OpenAI DevDay 2025 · Yahoo Finance
Sept 21, 2026 · Nearly 1 in 5 working-age people on Earth used generative AI (18.8%, Q2 2026). Microsoft AI Economy Institute
Film · 1 minute 35 seconds
pruf.net, from zero
Every app wants to know who you are. So we started from zero: a social network for humans only, where bots can’t get in and nothing about you is on file.
Every app wants to know who you are. Where you go. Who you know. What you like. And out there, it’s getting harder to tell who’s real.
So we started from zero. This is pruf.net. A social network for humans only. Bots can’t get in.
No email. No password. No face on file. Face ID stays on your phone.
Real moments. Photos people took themselves. No ads.
Your messages are encrypted end to end. Even pruf can’t read them. All your messages and files, encrypted at rest. In your pruf Vault, on your phone. No one can open it but you. Not the cloud. Not pruf.
Proof of humanity, without proof of identity. pruf.net. Join the public beta on iPhone.
The film is an illustration: the city, the door, the barrier and the cipher are pictures, not how pruf works inside, and the people in it are fictional. Face ID runs on your phone; PRUF keeps no face on file. For what PRUF does and doesn’t hold, see how privacy works.
Film · 58 seconds · cut paper
Real people, really private.
Our first film: a short cut-paper story about pruf.net.
Somewhere along the way, the internet filled up with bots… ads… and apps that want to know everything about you. Your email. Your password. Your face.
So we cut all that out.
This is pruf.net. A social network for real humans. One person, one phone. No email. No password. No face on file.
Connect with the people you actually know. Capture real moments, photos you took yourself, and share them with real people.
Your messages are end-to-end encrypted. We can’t read them either.
No ads. And bots? They can’t get in.
pruf.net. Proof of humanity, without proof of identity. Join the public beta on iPhone.
The film is an illustration. Face ID runs on your phone; PRUF keeps no face on file. For what PRUF does and doesn’t hold, see how privacy works.
Web series · Season 1
Why pruf.net
Pip loves sharing the little things. She doesn’t love being watched. A storybook in felt and stitch, one short episode at a time.
Vertical, made for phones held upright; captions are sewn into the picture.
Episode 1 · 42 seconds
Followed
Transcript
Every time Pip shares… the ads know too much, …and too many pictures never happened.
Pip, in a whisper: “I just wanted to share my sunrise.”
Kai: “Come post where I post.”
Real people. Photos they took themselves. No ads. Bots can’t get in.
Todd: “Hi, neighbor! I heard you like sunrises! I have all this cool sunrise gear to show you!”
To be continued…
The episode is an illustration. The old place is no real app, and Todd, the ads, the phones and the printer are not product mechanisms.
Episode 2 · 51 seconds
Nothing to hand over
Transcript
No email. No password. No face on file. Face ID, on her own phone.
The social network that doesn’t want to know your name.
Todd: “I’ll join pruf too, Pip!”
Todd: “Todd. A normal human.”
Bots can’t get in.
Todd: “Calling… everyone!”
To be continued…
The episode is an illustration. The old place is no real app, and Todd, the forms, the door, the phones and the masks are not product mechanisms.
Episode 3 · 51 seconds
Real people
Transcript
Todd: “Hi, neighbor!”
A thousand bots at the door.
Not one of them gets in.
Three hearts. Three real people.
Todd: “What are you posting, neighbor? I haven’t seen you online in a while!”
HQ: “Todd.” Todd: “On it!”
To be continued…
The episode is an illustration. The old place is no real app, and Todd, the Swarm, the door, the phones, the icons and the street are not product mechanisms.
Episode 4 · 50 seconds
Sealed
Transcript
HQ: “Todd. Get the messages.” Todd: “On it!”
Encrypted end to end.
Even pruf can’t read it.
Nana: “Hey T-Rexa, remind me: birthday tea, Saturday at four. Candles, and a cake.”
Todd: “Hi, neighbor! Candles?” Nana: “Funny. I only told the dinosaur.”
To be continued…
The episode is an illustration. The old place and T-Rexa are no real app or product, and Todd, the wire, the pole and the pad are not product mechanisms.
Episode 5 · 50 seconds
Nailed it
Transcript
Nana: “Oh, this is no good.”
Nana: “Why can’t I upload this to pruf?”
Pip: “Nana, pruf doesn’t allow uploads. Take your photo inside the app!”
Nana: “Okay. Fine.” … “Well. It’s brighter.”
June: “Nailed it 😂” Nana: “I did.”
Anyone can make perfect. Only you were there.
Meanwhile… Todd: “Duty calls!” … “Here’s your chance!”
Mr. Perfect: “I baked you something sweet! 🎂”
To be continued…
The episode is an illustration. The blue eye app, its AI Fix and T-Rexa are no real app or product; the gold frame, the filter strip and the camera are not the production app; Todd and Mr. Perfect are not product mechanisms.
Episode 6 · 49 seconds
Just one selfie
Transcript
Mr. Perfect: “No filter needed 😉”
Nana: “Send me an ugly selfie. I want to see you naturally.”
Mr. Perfect: “caught mid-sneeze 🤧” Nana: “That’s his ugly?”
Pip: “Show him how it’s done, Nana.”
Taken on pruf.net. By a human.
Nana: “Your turn.”
Meanwhile… Certified pruf photo · Taken by nanarose
Todd: “How are we gonna get on pruf?!”
Human not detected. Bots can’t get in.
Todd: “Abort! Abort! Abort!” · Mr. Perfect has left the chat.
Prove to anyone that these are your ugly selfies.
Pip: “Now that’s a face.”
The episode is an illustration. The blue eye app is no real app or product; Mr. Perfect’s photos are generated images of an original man who resembles no one; the seal, the gold frame and the public page’s look are not the production app, and the link’s code is made up. nanarose is a username she made up; pruf.net doesn’t know her name. Todd and Mr. Perfect are not product mechanisms.
Episode 7 · 52 seconds · the season finale, part 1
The Swap
Transcript
HQ: “Todd, get us her data.” Todd: “On it, boss.”
Pip unlocks her phone with her PIN. Todd reads it through his binoculars.
Todd: “Quickly, before she gets back!”
The PIN opens the phone. pruf.net opens only with Face ID: not a match.
Hacking storage… Open pruf Vault. Todd: “We’re in.” … “Encrypted.” … “All of it.”
In your pruf Vault, on your phone. No one can open it but you. Not the cloud. Not pruf.
Todd: “Straight to the source.”
To be continued…
The episode is an illustration. The phone, the cable, the hacking screen and the vault’s look are not the production app or how it works; Todd and HQ are not product mechanisms.
Episode 8 · 61 seconds · the season finale, part 2
Nothing on File
Transcript
HQ: “Find everything on her.” Todd: “Going in.”
He swings in through a window, past the lasers, opens a server panel and plugs straight into the drive. Todd: “Straight from the drive.”
Username: pip (public). Her public posts.
Todd: “No name… no email… no phone number…”
Name: none. Email: none. Phone number: none. Address: none. Face: not on file. Messages: encrypted. Advertising data: none.
Todd: “Nothing personal, boss! No name, no number!” HQ: “Then what do we sell?” Todd: “…Nothing?”
pruf. No record of who you are. Nothing to steal or sell.
HQ: “No data, no job. Todd, you’re recycled.”
WILL SPY FOR BYTES. Nana: “Poor thing.” Pip: “Smile!”
Captured by a human. End of Season 1.
The episode is an illustration. The tower, the lasers, the server panel, the drive and the record’s look are not PRUF’s systems or security; pip is a username she made up. Todd and HQ are not product mechanisms.
Web series · Privacy tips
Todd 2.0
Todd, a reformed ex-tracking bot, explains how apps and websites follow you, and how to stop it.
Vertical, made for phones held upright; captions are printed into the picture.
Episode 1 · 68 seconds
Why am I getting so many spam calls?
Transcript
Todd: “So many spam calls, all of a sudden? Your number gets around. …I helped.”
“Where I used to work, your number was a name tag. Email too. Same tag, so many doors. Made you easy to follow.”
“People-search sites sell reports: your address, your relatives. Number in, whole person out. Honestly? My best work. …Former work.”
“Action item: turn on your spam filter. iPhone: Settings, Apps, Phone, Spam. Android: Phone app, More, Settings, Caller ID and spam, Filter spam calls. Filtered calls don’t ring. Check voicemail.”
“Do Not Call? Scammers break the law. Robocall asks you to press 1, or say something? Don’t. It tells them your number works. Hang up.”
“Is there any app that doesn’t want your email and phone number?”
pruf.net. We don’t follow you.
Todd is a fictional ex-tracking bot, and the company he used to work for is fictional. His voice is synthetic. The phone number and email on screen are made up. The steps are for iPhone (iOS 27) and the Phone by Google app; on a Pixel, look under Spam and Call Screen, and other phone apps’ menus differ.
Todd: “Hi! It’s Todd, your unemployed spy bot, again. Did you know there’s a file on you? You didn’t write it. You didn’t ask for it. You’ve never seen it. Where I used to work, I wrote them.”
“Your ad profile. What you like. What you do. Apps, advertisers, data companies: they all use it to follow you, app to app.”
“In it: a number on your phone that advertisers can read. Your searches, when you’re signed in. Remember that time you searched garden gnomes at 2 a.m.? I do. Now gnomes haunt your feeds.”
“Your email. Many apps and sites turn it into the same code, so they can tell it’s you.”
“Quick fix. iPhone: Settings, Privacy and Security, Tracking. Off. Android: Settings, Google, All services, Ads, Delete advertising ID.”
“Is there a free social app that doesn’t keep an ad profile on you?”
pruf.net. We don’t follow you.
Todd is a fictional ex-tracking bot, and the company he used to work for is fictional. His voice is synthetic. The ad ID and email on screen are made up. On iPhone (iOS 27) the switch is Allow Apps to Request to Track. On Android, confirm the delete; menus differ, and on some phones it’s Settings, Privacy, Ads. You’ll still see ads. Fewer of them will know you.
Todd: “Hi! It’s Todd, your unemployed spy bot, again. Did you know you’re selling out your friends?”
“Give an app your contacts, and you hand over your friends’ names, numbers and emails. Where I used to work, one tap got us your whole address book. Including your dentist.”
“We suggested your friends to each other. Even in our other apps. We didn’t need your messages. Your contacts showed us who you know.”
“Don’t be that friend. Don’t share your contacts.”
“iPhone: Settings, Privacy and Security, Contacts, the app, Limited Access. Android: Settings, Security and Privacy, Privacy, Permission manager, Contacts, the app, Don’t allow.”
“Is there a free social app that never asks for your contacts?”
pruf.net. We don’t follow you.
Todd is a fictional ex-tracking bot, and the company he used to work for is fictional. His voice is synthetic. The names, numbers and emails on screen are made up. On iPhone (iOS 27), after Limited Access, tap Edit Selected Contacts and pick only the people you choose. On Android, menus differ; Settings, Apps, the app, Permissions also works. This stops new copying. It doesn’t delete contacts an app already uploaded: check that app’s own settings.
Todd: “Hi! It’s Todd, your unemployed spy bot, again. Is your phone listening to you, to sell you ads? It doesn’t have to.”
“You already told them. Where I used to work, we never needed your mic. You had thumbs.”
“Now, your free AI chatbot. The app store’s full of them. Talk to one, out loud even, and some use it for ads. One shows ads under its answers. Others keep your chats, and say people may review some. The rest? Who knows.”
“A free keyboard can send what you type. Who pays for free?”
“Your AI chatbot? Check its privacy settings. Turn off what you can.”
“Free app? Read its label: App Privacy, or Data safety.”
“Is there a free app that doesn’t listen to you, to sell you ads?”
pruf.net. We don’t follow you.
Todd is a fictional ex-tracking bot, and the company he used to work for is fictional. His voice is synthetic. The searches, chats and apps on screen are made up. In your AI chatbot’s privacy settings, turn off what you can: personalized ads, chat history or activity, and training on your chats. For a one-off question, use a temporary or incognito chat. Switching off doesn’t erase what’s already kept, and not every app has every switch. Before you download a free app, keyboard or AI, read its privacy label. iPhone: App Store, find the app, scroll to App Privacy, Data Used to Track You. Android: Google Play, find the app, scroll to Data safety. The app’s maker fills in its own label.