Rick Song

With how quickly fraud is evolving due to AI, online identity now collects more data than ever. Privacy feels like an inevitable sacrifice. We don’t believe it has to be, so we built Relay. Relay verifies you are human, but keeps your activity private. Persona never sees who you share this proof with or what you’re doing online, and the website you’re trying to verify with never sees any of your personal information. Identity and privacy don’t have to be a tradeoff.

When I was 13, I flew to Florida to hang out with someone I'd only ever played Warcraft 3 with online. I never met him in real life. His parents picked me up from the airport and I stayed for two weeks. We played games and hung out. It was a completely normal feeling at the time. Looking back, that was absurd. This was a person I knew entirely through a screen while playing an RPG. But it was also one of the more genuine friendships I had at that age. The internet didn't feel separate from real life. It felt like an extension of it. Back then, it kind of was. I think we can get back to that place.

Update: we detect streamer setups now!

3 of the worst parts of the internet: Slop, deceptive bots, and polarization. Right now, I feel we are in Slop Town. I estimate that 70%-80% of internet engagement and content right now is auto-generated. Engagement is manufactured. Discourse is polarized. Right now, we are not in a golden age of the internet. To fix this, we need to take identity seriously.

Verify you're you, share less. You shouldn't have to hand over your entire identity just to prove one thing about yourself. We've gotten used to sharing far more personal information than necessary online. Our work with @Ledger is a step toward changing that, giving people more control over what they reveal and what stays private.

No verified badge? ಠ_ಠ

The reason identity became such a big online conversation is actually pretty simple: everything moved online. You used to have to physically show up to prove who you were, drive to the bank, walk into the restaurant, stand in line at the box office… All of it required your actual self to be somewhere. Now none of those key moments of verification require you to leave your couch (or your bed). Identity just had to catch up to where the interaction moved.

Greg is dope

There’s a lot of talk about AI safety and not enough about AI accessibility. There's so much talk about p(doom), AI hacking into everything, and the end of the world. But for the average person, I think a much more immediate anxiety is the feeling of being left behind. They hear that AI is transforming everything, try it themselves, watch an agent fail to buy their groceries, and wonder, "What am I missing?" It creates this strange gap where people are constantly told AI is incredibly powerful, but don’t experience that power themselves. It feels like everyone else has access to some incredible technology that they somehow missed. Safety matters, but I wish there was more discussion about how we can build things to make sure everyone can actually benefit from AI.

We pointed Opus 5.5 at one of our research codebases and told it to visualize how it works. If you ever wonder how @withpersona thinks about deepfakes and fraud, here's a sneak peek at what's happening under the hood.

The internet needs proof of humanness more than ever. We have a unique opportunity to build it with privacy at the core. As bots become harder to stop and identity theft grows more sophisticated, platforms increasingly need stronger ways to verify who’s real. But stronger verification shouldn’t have to mean collecting more personal data. We should be asking not just what data is needed, but also what data isn’t. That’s what the team at @withpersona is working toward with Relay. Relay uses Privacy Pass, an open IETF standard, to let organizations verify claims like humanness or age without learning your identity, all while preserving the best-in-class accuracy of Persona’s platform. Proud of the team for the work on this one. Excited to see where it goes

It’s crazy that AI can just do this now…

Who could have guessed that fraud detection data is kind of beautiful? Looks more like a topography map or mitochondria than a bunch of fraud clusters.

Our team at @withpersona has spent years working on human verification. We've built one of the most robust systems to prove humanness, but as identity becomes ever more critical, we need to rethink how it works. The future shouldn't be scanning your face multiple times a day to use the internet. There’s still a lot of work for us to do here, but we are excited for what's next.

Our team at @withpersona has learned a lot about identity fraud from our work helping stop some of the most sophisticated fraud attacks. Really excited about about the work they're doing with @UBuffalo researchers to map the landscape and create an interactive resource and open dataset that everyone can use to improve our collective efforts to protect people from identity theft and fraud. I recently wrote about how identity attacks are changing. This is one way we're turning what we see in the real world into something researchers and fraud teams can study, test against, and build on. The Hugging Face dataset gives researchers synthetic and recreated examples of real-world attacks to train and test models, based entirely on patterns we've encountered in the wild. Really proud of the @withpersona Research team, @iamcharbles, Saatvik Billa, Hongkai Pan