
Investing my time in humble offerings that makes your life a little more private. Building @onflowxyz & @protectmyface.
Europe
Joined March 2024
- Tweets349
- Following404
- Followers3.8K
- Likes5.6K
Pinned Tweet
Proud to announce our first retail product. Go check it out. Find yourself anywhere online with maximal privacy and get it taken down immediately.
Get notified whenever someone posts your face on the internet without your knowledge. And get the option to take it down.
Today I'm launching a passion project: protectmyface.org
It's two things in one: continuously scan the internet for people posting your face, alerts you when it shows up, and the ability to take down those photos with a simple button. All in one subscription.
Why would I need this?
Curiosity: Most people would like to know if someone is posting their photos on random sites without their knowledge.
Digital hygiene: The ability to take down those photos lets you clean your digital footprint, especially for photos you don't like.
Bullying: There's a huge uptick in image-based bullying, especially with AI edits using someone's real face.
Peace of mind: even if you never exercise the takedown option, just knowing it's one click away is genuinely reassuring.
How does it work?
You upload a simple selfie and our system creates a private vector out of it, which is just a list of numbers. That vector is your unique math.
The vector can recognize when another face produces the same numbers.
We scan the internet comparing your vector against images we find online, focusing on the high-risk surfaces where image abuse actually clusters: image boards like 4chan, random public hate forums, and the broader public web.
When we detect a match we notify you through a simple email.
Like "hey, your face appeared on 4chan, a random forum, or a profile that isn't yours. Here's the link", and a convenient button to take it down using applicable laws that fit that case.
All notifications are through email so you don't have to keep checking the website.
Our scanner is continuous and always growing. It's not a one-time check. You sign up once and it keeps working in the background. have the peace of mind that if someone posts your face 9 months from now on a site we just added to our coverage, you'll still get notified. Coverage compounds.
CAN THIS BE USED TO STALK SOMEONE!?
No. ProtectMyFace is designed from the ground up to ONLY protect your own face. We use multiple verification methods to ensure the face being uploaded belongs to the account holder, including liveness checks and ID verification on flagged accounts.
Uploading someone else's face violates our terms of service and will result in account termination. We take abuse prevention seriously and actively monitor for misuse.
We also don't immediately deliver search results until a few days after account creation to make sure the account holder is protecting their own face and not using it to search for someone else's.
Our commitment to safety is a big differentiator between us and other services that let anyone search for photos that aren't theirs with zero safeguards.
Do you have experience building something this large? Is it secure?
ProtectMyFace isn't just me. It's built by Sundial, a team of privacy and identity engineers with years of experience building products where data security is non-negotiable. Our ongoing projects include Onflow and other identity products that handle critical data at scale. We bring the same standards to this project.
You scan the internet? Isn't that an almost impossible task?
It is. But we use clever smart crawling methods that target the most high-risk clusters of the internet and common places where image-based abuse happens most often. We also scan the general public web slowly as our infrastructure grows. So we're only getting better from here.
You can also submit links to photos you want taken down directly through your dashboard. So it's not only for photos our internet scanner finds. If you already know about a post using your face, you can hand us the link and we'll handle the takedown for you.
SIGNAL retweeted
We've added a convenient one-button way to file Take It Down Act requests on protectmyface.org if you ever need it. It covers AI edits as well.
#TAKEITDOWNAct requires covered platforms to give victims a way to quickly remove intimate images and videos shared without their consent—and the platforms must do so within 48 hours of a valid request. FTC collects reports about violations of the law at TakeItDown.ftc.gov
Platforms only listen when there's money on the line. If they try to ignore reports now the FTC fine is up to $53,088 per violation.
Don’t know what images are out there of you? Where your face appear and what people might have done with them.
@protectmyface universally solved this and removed material about you in the most privacy-respecting way.
Read more on the website for more info.
Privacy for good (and free).
Replying to @arc4g
One more thing. If you're dealing with a serious case, harassment or bullying using your photos, please reach out. We'll help you for free.
SIGNAL retweeted
When I first started uploading photos of myself to the internet, they would often get photoshopped into gay porn
I didn’t particularly have a problem with this, but people would send it to my mom and it pissed her off
Arc designed this product to help identify gay porn of my face and DMCA it off the internet before my parents could see
Later we discovered this also works really well for women who want to do the same thing for any AI generated imagery of themselves, which makes a lot more sense as a business strategy
Get notified whenever someone posts your face on the internet without your knowledge. And get the option to take it down.
Today I'm launching a passion project: protectmyface.org
It's two things in one: continuously scan the internet for people posting your face, alerts you when it shows up, and the ability to take down those photos with a simple button. All in one subscription.
Why would I need this?
Curiosity: Most people would like to know if someone is posting their photos on random sites without their knowledge.
Digital hygiene: The ability to take down those photos lets you clean your digital footprint, especially for photos you don't like.
Bullying: There's a huge uptick in image-based bullying, especially with AI edits using someone's real face.
Peace of mind: even if you never exercise the takedown option, just knowing it's one click away is genuinely reassuring.
How does it work?
You upload a simple selfie and our system creates a private vector out of it, which is just a list of numbers. That vector is your unique math.
The vector can recognize when another face produces the same numbers.
We scan the internet comparing your vector against images we find online, focusing on the high-risk surfaces where image abuse actually clusters: image boards like 4chan, random public hate forums, and the broader public web.
When we detect a match we notify you through a simple email.
Like "hey, your face appeared on 4chan, a random forum, or a profile that isn't yours. Here's the link", and a convenient button to take it down using applicable laws that fit that case.
All notifications are through email so you don't have to keep checking the website.
Our scanner is continuous and always growing. It's not a one-time check. You sign up once and it keeps working in the background. have the peace of mind that if someone posts your face 9 months from now on a site we just added to our coverage, you'll still get notified. Coverage compounds.
CAN THIS BE USED TO STALK SOMEONE!?
No. ProtectMyFace is designed from the ground up to ONLY protect your own face. We use multiple verification methods to ensure the face being uploaded belongs to the account holder, including liveness checks and ID verification on flagged accounts.
Uploading someone else's face violates our terms of service and will result in account termination. We take abuse prevention seriously and actively monitor for misuse.
We also don't immediately deliver search results until a few days after account creation to make sure the account holder is protecting their own face and not using it to search for someone else's.
Our commitment to safety is a big differentiator between us and other services that let anyone search for photos that aren't theirs with zero safeguards.
Do you have experience building something this large? Is it secure?
ProtectMyFace isn't just me. It's built by Sundial, a team of privacy and identity engineers with years of experience building products where data security is non-negotiable. Our ongoing projects include Onflow and other identity products that handle critical data at scale. We bring the same standards to this project.
You scan the internet? Isn't that an almost impossible task?
It is. But we use clever smart crawling methods that target the most high-risk clusters of the internet and common places where image-based abuse happens most often. We also scan the general public web slowly as our infrastructure grows. So we're only getting better from here.
You can also submit links to photos you want taken down directly through your dashboard. So it's not only for photos our internet scanner finds. If you already know about a post using your face, you can hand us the link and we'll handle the takedown for you.
yeah ICANN is known for their strict enforcement of false names on .org tlds.
he probably literally just put in “satoshi nakamoto” for the name.
Satoshi Nakamoto registered e-cash.org on July 20, 2008, a month before registering bitcoin.org
He's never given up the domain name and it's currently registered with Dynadot.
Meaning the owners of Dynadot literally know who Satoshi Nakamoto is.
Readers added context they thought people might want to know
The e-cash.org domain was first registered in 2002, has expired multiple times, and changed registrars and owners, including acquisition by Dynadot in 2019 after another expiration, so it was not registered by Satoshi in 2008 and continuously held since. x.com/jeftovic/statu… x.com/0xLimez/status…
You’re either pro-privacy or an ontologically flawed soul.
There is no middle ground, and I’m tired of pretending otherwise.
deployed a small website in light of the recent google paper on the quantum attacks on bitcoin and ethereum with post quantum tech.
quantumbags.com
it explains what it is, and what it isn't (there's also an article incoming), and you can verify the ZKP they provided.
Tim Cook spending ZERO on AI, waiting for the bubble to pop so he can buy it at comical prices, while he sells more iPhones than God.
One of the best to ever do it, but my god do I wish I didn't have to annunciate my words screaming to turn on the lights with Siri.
your entire net worth in a browser extension with LLM zero days dropping daily. no way this goes wrong for you.
build beautiful software.
we launch 2 things this month:
- onflow (max privacy respecting identity/kyc) SDK
- a surprise privacy service you’ll want to use (more info soon)
pro-humanity tools without the asterisk is a moat (and infinitely more rewarding).
This month will be the culmination of >1.5 years of work. 10 people working weekends, waking up in the morning and working into the night on a project, for over a year and a half. Everything from software engineers, cryptographers & BDs.
One of the most incredible journeys I’ve had the pleasure of being a part of.
We’ve landed some incredible partnerships, and are going to be working with a ton of great startups when @onflowxyz launches this month.
Building useful, and perhaps most importantly, privacy enhancing tools for developers and users is my dream “job” (due to fortunate circumstances, many executives at Sundial doesn’t take a salary). Seeing it finally materialize and potentially outgrow what I dreamed possible is surreal.
I’ll keep you updated on the journey here going forward. I’d also like to thank those who believed in our vision, and those who backed it to make our world a bit more private (which has made Onflow an increasingly relevant tool in the age of age verification).
Thank you. And I hope you go find a team you love and make your values a reality that people appreciate.
Seems like there's an insane amount of demand for truly private email. (Read entire post)
A few questions for those who want a better Proton:
1. Are you happy with the UX?
2. Anything else you'd like to see?
3. Premium plans; what should they include?
4. Credit cards (payment info deleted) + XMR?
5. Do you prefer a web client like Proton or a desktop client like Outlook (or both?)
And finally, if you have any exceptional names with the .com being available, you'll get the highest tier for free, forever (capped storage, custom domains, etc.).
We're currently reviewing the cryptographic soundness of the protocol we've come up with, so far it's looking very promising, and we'll finally have truly private email, where there's zero trust in us (the service provider) saving your data (as we can prove via ZKPs we a single instance never saw it, same with metadata).
This also means if your account gets hacked, it's pointless since they won't be able to decrypt your email.
Here's what we can guarantee: we can never see what you send and receive, but obviously due to this being standard email if you send it out to another email service provider like Outlook, they will be able to see the content.
But email being the single point of failure for all of your accounts getting compromised will be a thing of the past.
Glad to see the spirit and demand for true privacy services is alive and well. Share your feedback, and we'll note you down as early adopters.
I've been vocally critic of @ProtonMail.
Exposure of them doxing a user today made me rethink how we can use the leaps in cryptography to achieve truly private email without changing SMTP.
We did it. Provably private mail is coming.
No more single point of failure. 🫡💛
Our mail service, Ironmail, has been officially announced.
Check this out if you want to reserve a handle and get 2 months free of the Enthusiast plan (no card required):
xitter.cf/SIGNAL_RETURN/status/2…
Announcement: our TEE/ZK/MPC privacy email service is will be called Iron Mail.
Our current estimate for a release date is start of Q3 2026 (roughly a 4 month development time).
If you read the article below, and comment we'll award you with a perk (read on).
A lot of people boldly claim "SMTP is old, therefore privacy is impossible". That's fair, if you say it with an open mind. What is also true is that multi-party computation and zero-knowledge proofs have come a long way, and so has TEEs.
By utilizing a novel stack of these technologies (a stack we call IRON), we can ensure that even if served with a subpeona, we cannot capture data while at rest, while in transit (both inbound and outbound), this includes metadata.
We will have free options that give you up to 1 GB of storage, as many folders as you need. Pinned mails, and so much more.
While I can't go fully into detail on the exact mechanisms of how the IRON stack works yet to achieve in-platform privacy soundness, you'll be getting it near launch to verify for yourself exactly how we make sure that you can still receive, send and store emails with no option to for the service provider (us), to read your emails. Period.
Iron Mail is developed for everyone to use. We'll be doing a public ceremony to get rid of so called "toxic waste" (generating a CRS which anyone who's verified can participate in) as the entire service relies on zero-knowledge proofs that every user can verify for their email dispatch that:
1) inbound is decrypted within a TEE edge server.
2) the plaintext was directly sent to an MPC council.
3) mail routing and attribution was done correctly.
4) SMTP-specific things such as DKIM verification.
5) outbound is jointly broadcast through MPC collaboration with the final email not being visible to the service provider.
All zero-knowledge circuits will be released, after they've been audited by several parties, you can audit them yourself. This means, there is no uncertainty on whether we have hidden log infrastructure, since that would be either a) tampering with the process, which would invalidate the proof chain or b) be outright impossible once it's being processed by the MPC committee.
Setting expectations
1. Referrals
On referrals. We will have them. If we sponsor influencers, they will not be given a script, they can give their honest opinion. We will only be giving sponsorships to niche technical channels who actually knows what they're vouching for. Ironmail is mostly meant to be deployed for institutional and governmental security.
Running a permanent free service for the public is purely in the interest of promoting good opsec and giving people access to a new innovation in a space where there hasn't been much for years now, despite annual earnings in excess of $100M+ (for Proton Privacy in 2024, per Zoominfo).
We might also give referral code to tech enthusiasts who write blogs or has an audience, as these are the people who can accurately assess the veracity of the claims made. This has remained our biggest issue with companies like Proton: the marketing.
We've seen them be subversive or sometimes, what can easily be suspected to be, intentionally misleading.
2. Defending users
Based out of Delaware, US., jurisdictionally, we'll stand up for every subpoena, and we will always carry an up-to-date warrant canary in the footer of the website that we urge you to check (or setup an automated curl for), to ensure the company is still in our control.
Signing up will be through 3 options:
1. A proof of work puzzle that takes between 5-10 minutes to solve on most processors.
2. An SMS message (after which that number is marked "used", but is not associated with your email).
3. By using @onflowxyz, zero-knowledge based proof of uniqueness using an ICAO 9303 document (i.e. a passport, or ID card, none of the info is transmitted to us, only a zero-knowledge proof, which is merely a mathematical proof of the validity of information, not any of the actual information).
Any request for user info, be it IP address, email, contents of email, phone number, name or setting up future snooping is impossible within the IRON protocol. But again, we always urge you to check the warrant canary.
Accessing the Site
We will offer both a web UI and a fork of Thunderbird to access the Ironmail. We will also have endpoints setup for TOR and I2P (though experiences may be slightly degraded in terms of smoothness due to JS blocking) if you do not want your ISP to know you're using privacy-respecting services.
Users can sign up on a VPN (we suggest Mullvad), on TOR or via I2P.
You can pay via credit card where we give you the option to pay via credit/debit card (but, it is not recurring, and all payment data is deleted after 20 days), you can decide if you want to renew.
You can also pay using cryptocurrency, including Monero and Zcash shielded transactions.
We want to keep as little information about you as possible. We do not store which transaction hash was associated with your purchase after the sufficient number of verifications.
Pricing
The service, as mentioned, is meant for governmental and institutional usage in private deployments, however for our public-facing site, which can also be used by organizations of any size, the pricing will be similar to that of Proton, simply due to a much larger computational overhead than something like Proton (MPC and TEE isn't cheap in overhead, neither is ZK).
The service is 100% free for everyone who doesn't wish to use an extended feature set, and comes with half a gigabyte of storage by for the free tier.
Political affilations
We are not aligned with any party or ideology as a company, and our team rarely if ever discusses politics, we discuss privacy.
Launch
Due to the complex nature of the engineering the protocol to be production confident, then the auditing process, as well as stress testing, fuzzing and vulnerability testing along the way, combined with a UX we hope can rival the best, launch is set to be in approximately 4-5 months.
The Prize
Thank you for reading this. I've garnered a ton of support personally from a privacy-conscious crowd for talking about privacy pretty much everyday, from interesting papers to Proton to zero-knowledge and whatever else is on my mind.
Sam Bent makes amazing videos, and apparently also, amazing blog posts (and hopefully he'll also make a video on the blog post he made).
If you share what you liked most about his article, and why privacy-preserving emails matter to you we will reserve a handle for you, plus give you 2 months of the Enthusiast tier for free (use your own domains, more storage, aliases and unlimited email addresses).
There is no upfront cost, we will simply send a code for you to enter once you sign up, which will also verify that you're trying to register the handle you reserved (some handles go quickly like first names, short words, etc.).
Thank you
For continuing the fight for privacy. We hope you're excited about the innovation we're trying to bring to this sector that has been very good at twisting words, using unclear marketing, only speaking up when caught and generally has tarnished the word 'privacy' for the average person in what they should expect.
The more services LARP as true privacy-oriented services, that then turn out to be something different in highly publicized events, hurts the reputation of everyone else trying to build for privacy.
And finally, good article Sam. Keep writing these.
Announcement: our TEE/ZK/MPC privacy email service is will be called Iron Mail.
Our current estimate for a release date is start of Q3 2026 (roughly a 4 month development time).
If you read the article below, and comment we'll award you with a perk (read on).
A lot of people boldly claim "SMTP is old, therefore privacy is impossible". That's fair, if you say it with an open mind. What is also true is that multi-party computation and zero-knowledge proofs have come a long way, and so has TEEs.
By utilizing a novel stack of these technologies (a stack we call IRON), we can ensure that even if served with a subpeona, we cannot capture data while at rest, while in transit (both inbound and outbound), this includes metadata.
We will have free options that give you up to 1 GB of storage, as many folders as you need. Pinned mails, and so much more.
While I can't go fully into detail on the exact mechanisms of how the IRON stack works yet to achieve in-platform privacy soundness, you'll be getting it near launch to verify for yourself exactly how we make sure that you can still receive, send and store emails with no option to for the service provider (us), to read your emails. Period.
Iron Mail is developed for everyone to use. We'll be doing a public ceremony to get rid of so called "toxic waste" (generating a CRS which anyone who's verified can participate in) as the entire service relies on zero-knowledge proofs that every user can verify for their email dispatch that:
1) inbound is decrypted within a TEE edge server.
2) the plaintext was directly sent to an MPC council.
3) mail routing and attribution was done correctly.
4) SMTP-specific things such as DKIM verification.
5) outbound is jointly broadcast through MPC collaboration with the final email not being visible to the service provider.
All zero-knowledge circuits will be released, after they've been audited by several parties, you can audit them yourself. This means, there is no uncertainty on whether we have hidden log infrastructure, since that would be either a) tampering with the process, which would invalidate the proof chain or b) be outright impossible once it's being processed by the MPC committee.
Setting expectations
1. Referrals
On referrals. We will have them. If we sponsor influencers, they will not be given a script, they can give their honest opinion. We will only be giving sponsorships to niche technical channels who actually knows what they're vouching for. Ironmail is mostly meant to be deployed for institutional and governmental security.
Running a permanent free service for the public is purely in the interest of promoting good opsec and giving people access to a new innovation in a space where there hasn't been much for years now, despite annual earnings in excess of $100M+ (for Proton Privacy in 2024, per Zoominfo).
We might also give referral code to tech enthusiasts who write blogs or has an audience, as these are the people who can accurately assess the veracity of the claims made. This has remained our biggest issue with companies like Proton: the marketing.
We've seen them be subversive or sometimes, what can easily be suspected to be, intentionally misleading.
2. Defending users
Based out of Delaware, US., jurisdictionally, we'll stand up for every subpoena, and we will always carry an up-to-date warrant canary in the footer of the website that we urge you to check (or setup an automated curl for), to ensure the company is still in our control.
Signing up will be through 3 options:
1. A proof of work puzzle that takes between 5-10 minutes to solve on most processors.
2. An SMS message (after which that number is marked "used", but is not associated with your email).
3. By using @onflowxyz, zero-knowledge based proof of uniqueness using an ICAO 9303 document (i.e. a passport, or ID card, none of the info is transmitted to us, only a zero-knowledge proof, which is merely a mathematical proof of the validity of information, not any of the actual information).
Any request for user info, be it IP address, email, contents of email, phone number, name or setting up future snooping is impossible within the IRON protocol. But again, we always urge you to check the warrant canary.
Accessing the Site
We will offer both a web UI and a fork of Thunderbird to access the Ironmail. We will also have endpoints setup for TOR and I2P (though experiences may be slightly degraded in terms of smoothness due to JS blocking) if you do not want your ISP to know you're using privacy-respecting services.
Users can sign up on a VPN (we suggest Mullvad), on TOR or via I2P.
You can pay via credit card where we give you the option to pay via credit/debit card (but, it is not recurring, and all payment data is deleted after 20 days), you can decide if you want to renew.
You can also pay using cryptocurrency, including Monero and Zcash shielded transactions.
We want to keep as little information about you as possible. We do not store which transaction hash was associated with your purchase after the sufficient number of verifications.
Pricing
The service, as mentioned, is meant for governmental and institutional usage in private deployments, however for our public-facing site, which can also be used by organizations of any size, the pricing will be similar to that of Proton, simply due to a much larger computational overhead than something like Proton (MPC and TEE isn't cheap in overhead, neither is ZK).
The service is 100% free for everyone who doesn't wish to use an extended feature set, and comes with half a gigabyte of storage by for the free tier.
Political affilations
We are not aligned with any party or ideology as a company, and our team rarely if ever discusses politics, we discuss privacy.
Launch
Due to the complex nature of the engineering the protocol to be production confident, then the auditing process, as well as stress testing, fuzzing and vulnerability testing along the way, combined with a UX we hope can rival the best, launch is set to be in approximately 4-5 months.
The Prize
Thank you for reading this. I've garnered a ton of support personally from a privacy-conscious crowd for talking about privacy pretty much everyday, from interesting papers to Proton to zero-knowledge and whatever else is on my mind.
Sam Bent makes amazing videos, and apparently also, amazing blog posts (and hopefully he'll also make a video on the blog post he made).
If you share what you liked most about his article, and why privacy-preserving emails matter to you we will reserve a handle for you, plus give you 2 months of the Enthusiast tier for free (use your own domains, more storage, aliases and unlimited email addresses).
There is no upfront cost, we will simply send a code for you to enter once you sign up, which will also verify that you're trying to register the handle you reserved (some handles go quickly like first names, short words, etc.).
Thank you
For continuing the fight for privacy. We hope you're excited about the innovation we're trying to bring to this sector that has been very good at twisting words, using unclear marketing, only speaking up when caught and generally has tarnished the word 'privacy' for the average person in what they should expect.
The more services LARP as true privacy-oriented services, that then turn out to be something different in highly publicized events, hurts the reputation of everyone else trying to build for privacy.
And finally, good article Sam. Keep writing these.





















