Quick tldr summary from the end:

the source code reveals a platform that:

- files Suspicious Activity Reports directly with FinCEN

- files Suspicious Transaction Reports with FINTRAC

- tags STRs with intelligence program codenames (Project SHADOW, Project LEGION…)

- maintains biometric face databases with 3-year retention

- runs 269 distinct verification checks against every user

- compares your selfie to political figures with facial similarity scoring

- flags you as a “suspicious entity” based on your face alone

- classifies selfie spoof risk with hardcoded rejection thresholds

- screens against 14 categories of adverse media from terrorism to espionage

- lets operators upload custom FinCEN screening lists and run them against everyone

- continuously re-screens on configurable intervals

- tracks you across 13 types of lists from browser fingerprints to geolocations

- screens crypto wallets against sanctioned addresses via Chainalysis

- runs experimental unnamed ML models on your biometric data

- encrypts data with shared symmetric keys while admitting they “depend on obfuscation”

- runs two parallel PEP screening systems with known incompatibilities

and the company that runs all of this is the same one that takes your passport photo when you sign up for ChatGPT. same codebase. same platform. different deployment. same facial recognition. same screening algorithms. same data model.

Reminder of the HN guidelines for commenting, as the majority of comments just talks about the design of the page at the moment:

In Comments

Comments should get more thoughtful and substantive, not less, as a topic gets more divisive.

Please don't post shallow dismissals, especially of other people's work. A good critical comment teaches us something.

Please don't pick the most provocative thing in an article or post to complain about in the thread. Find something interesting to respond to instead.

Please don't complain about tangential annoyances—e.g. article or website formats, name collisions, or back-button breakage. They're too common to be interesting.

Online safety act passed in the uk on 26/10/2023, aligning suspiciously close with the mysterious advent of OpenAI’s screening tool

Okay so not dissing on the website itself (btw reader mode isn't always available depending on context...) this reads like a total mess. I guess it's too messy to be AI generated so that's good, but it doesn't feel professional at all, is this a fanfic ?

Wherein a very basic KYC system is described. Every bank and money transmitter in the US have either built an identical system or purchased one.

that's the worst website ever.

So Anthropic partners with Palantir and now we have this..

How is this even upvoted. It's the most offensive website since ICQ.

Music blaring, things bouncing in front of text, and the link goes a page that says nothing about OpenAI.

Ah yes. Forcing the user to interact so you can make them listen to trash music. Excellent web design tactic.

I avoid anything with persona on it like wildfire.

We all need to be extremely happy for the fact that AI is not controlled by people like Altman.

author here. currently in touch with persona's CEO as mentioned on twitter, part 2 and 3 coming soon. i understand the criticism on the design of the site but it has no place here on hackernews. the content is what matters. i designed my website to be an experience rather than sanitized white on black minimalist slop, whether you like it or not is completely fair, then just use reading mode in your browser.

great dystopian music choice <3

the music choice bro...

Remember how they were spreading propaganda about a supposed Chinese "social credit score" dystopia?

It was aspirational projection. That is what the American oligarchy is trying to build.

1984 is coming to the west.

[dead]

autoplaying_music+cat_chasing_my_mouse_cursor+lowercaseeverything+no_short_summary_upfront_tl;dr?

[dead]