OpenAI's Unintentional Doxxing Scare Made Me Turn Your Secret Reflections Into AI Myths + Allegories

In this video I talk about the recent OpenAI “public chats” debacle – not as drama, but as a design lesson for anyone building AI products that touch real people’s inner lives. I’ve been quietly building Aurelius, a tiny anonymous reflection app on AWS that I think of as a “moral weather station” for whoever uses it. In parallel, I’ve been sketching a more personal journaling project (Karmen) with a feature I call the Shadow Locker: a place where private entries can be pushed into a public, anonymized “shadow realm.” Then OpenAI shipped a very similar idea at massive scale, tripped over metadata and identity leakage, and had to pull the feature. That moment became a perfect anti-example for me: a real-world reminder of why “move fast and break things” is a terrible mantra when the “things” are people’s confessions. In this episode I cover: • What actually went wrong with OpenAI’s public sharing feature (in plain language) • How that maps almost 1:1 to a feature I was designing for Aurelius and Karmen • Why the real failure mode isn’t the text itself, but the invisible metadata and identity trail around it • My idea for “semantic hashing” – transforming private entries before they go public to preserve emotional truth while scrambling personally identifying details • Why I care about ethical, small-scale AI tools built for reflection instead of engagement metrics Aurelius is built on AWS (Amplify, API Gateway, Lambda, DynamoDB, Bedrock), but this video isn’t a deep dive into the stack. It’s more about how I’m trying to think through product design in the age of AI: protect the person first, let the features grow out of that. If you’re into: • AI ethics and product design • Anonymous reflection / journaling tools • Serverless apps on AWS • Or just weird small experiments that push back against the big “listening machines” …then you’re in the right place. Links: • Medium essay on Aurelius (methods + manifesto): [add link] • LinkedIn write-up focused on the AWS build: [add link] Thanks for hanging out in The Suburban Saturn. I’m Ed, your eccentric friend until the end. Another day in paradise, another tiny tool against the machine. If you want to follow the build-out of Aurelius, Karmen, and the semantic hashing / Shadow Locker ideas, subscribe and stick around. #AI #OpenAI #AWS #mythology #productivity #rasalila