Most people have a ready answer when privacy comes up: my life is boring, so who would want my data? Work, school run, gym, groceries, dinner, bed. There's nothing to steal.

It's a reasonable instinct, and it's wrong. That was where the first episode of Axon Unscripted ended up, in a conversation with Vini Barbosa. Vini started in crypto as a journalist covering hacks, ended up being invited into war rooms when projects were exploited, and now helps teams with security, documentation and open-source work through thecoding.dev. He has seen a lot of incidents up close, and most of them had nothing to do with the victim being rich or interesting.


Being boring doesn't make you safe

Whether your data is valuable isn't really about how interesting your life is. What matters is what someone can do with it once they have it.

The most common use is ordinary. A scammer who knows your address, your ID number and your mother's maiden name can write you a message that feels like it comes from someone you trust. That trust is the whole attack. It gets you to click the link, install the file or type in your card number. You don't need to be wealthy to be worth that effort. If you have a bank account and something to lose, you qualify, and that covers almost everyone.

Money is only one dimension. Health data is valuable whoever you are and whatever you earn. A diagnosis, a prescription or a pattern in your sleep can be used to impersonate a doctor, to pressure you, or simply to be sold on. Vini summed it up well: it doesn't matter until it does. By the time it matters, it's too late to take it back.


We hand it over without noticing

The uncomfortable part is how much of this we give away voluntarily, and AI has made it worse.

Take one everyday request. You upload a card statement to a chatbot and ask where your money is going. That single file holds your home address, your bank, your balance, where you shop, when you shop, and anything you'd rather nobody saw. Multiply that by everything people now paste into AI tools: medical results, contracts, private code, family photos. The companies running these models end up holding enormous, detailed profiles, which makes them very attractive targets. Even if a provider never sells a byte, one breach is enough.

Social media does the same thing more slowly. Every photo of your kids, your street or your front door adds to a picture of your life that never gets deleted. The internet doesn't forget, and it's very good at connecting dots.


Open source matters more in the AI era

So who can you trust with your data? The honest answer is whoever lets you check.

Terms and conditions don't really let you check. They tell you what a company says it does. You rarely learn what actually happens unless a lawsuit forces it into the open, and those usually end in settlements that are small change for companies that size. Christian has a version of this story most people will recognise: mention holiday plans to a friend in a private chat, and the ads start arriving within the hour. Maybe there's an innocent explanation. You just have no way of knowing.

Open source changes that. Vini's view is blunt: if something isn't open source, it isn't really decentralized. For years the catch was that only programmers could read the code, so openness was a nice principle most people couldn't use. AI has quietly removed that barrier. Today you can point an AI agent at a codebase and ask, in plain English, what it does with your data. Checking used to be a specialist skill, and now it takes a prompt.

Cost is part of this too. Many of today's AI subscriptions are heavily subsidised, and nobody outside the big labs knows what the models actually cost to run. Christian tracks his own usage and regularly burns through a month's subscription worth of compute in a single day. That won't last. When the subsidies end, the people who handed over the most data will have the hardest time leaving. Open models, which Vini uses day to day and finds both capable and much cheaper, are the exit door.


Privacy tech helps, but it isn't magic

The industry's answer to "can I trust the server?" is increasingly hardware that keeps data sealed even from the company running the machine, with proofs that anyone can verify. It's a real improvement on sending everything to an ordinary server, and Vini rates it.

He was also careful to add the caveat that marketing usually leaves out: hardware and software both have bugs. If an attacker finds the right weakness, sealed data can still be read. Protection doesn't make anything invulnerable. The safest setup is still the one where your data never leaves a machine you control.


Three habits worth picking up

None of this requires becoming a security expert. Vini's starting list is short:

  1. Share less, and share more thoughtfully. Before you post, ask what the picture gives away. Take the selfie facing away from your front door.
  2. Use a VPN you pay for. Your IP address reveals your rough location and follows you across every site. Avoid free VPN apps, though. Running a VPN costs money, so a free one is earning it somewhere, often from your traffic. Christian uses Mullvad. No sponsorship, just a recommendation.
  3. Know what you install. This one is new. AI coding tools can now install packages on your behalf, sometimes in modes that skip permission checks entirely. That's fine on a spare machine set up for it. On the laptop that holds your whole life, it's an open door.


Code is law, so write better code

The episode closed on an idea Christian has been turning over since law school, where his thesis asked whether a smart contract could be enforced like a traditional one. The Mango Markets case still divides people: the trader behind it argued he simply understood the protocol better than the people who built it. In traditional contracts, finding a loophole is often just called good lawyering.

Vini's answer is the right one. Code is law, and some law is bad. Society keeps rewriting bad law, and builders should keep rewriting bad code. Security never reaches a finished state. Someone has to keep improving it.


Why we started here

Axon Unscripted is a series of unscripted conversations with people working across AI, decentralized science and biotech. We opened with privacy because it sits under everything else we care about: who holds your data, and whether you get a say in what happens to it.

It's also the idea behind AxonLife, which is being designed to keep health data under the control of the person it belongs to.

Watch the full conversation: https://youtu.be/GKbGy37BRRw

Follow Vini on X: @vinibarbosabr