Security, Privacy, and Philosophy: The New "Trends" of IT

This is just a fun story. Well, mostly.

I grew up with the screeching sound of dial-up internet and a phone line I had to share with my grandma. When the internet finally became fast and cheap, we all did the same thing: we uploaded everything. Photos, locations, opinions, breakfast. The CEO of Sun Microsystems told us back in 1999: “You have zero privacy anyway. Get over it.” Mark Zuckerberg explained in 2010 that privacy was simply no longer a social norm. And honestly? We believed them. Data was “the new oil,” sharing was caring, and the industry’s official motto was “move fast and break things.”

I was there for all of it. I reinstalled Windows for friends, found creative ways around the settings their dads had configured, and later fell in love with Linux and assembly language. The internet felt like a candy store. Nobody read the price tags.

Then the bill arrived.

The bill

Not a metaphorical bill. An actual, itemized invoice.

A data breach in the US now costs an average of $10.22 million - a record, in case anyone was collecting achievements. European regulators have handed out more than 7 billion euros in GDPR fines. One data broker, National Public Data, leaked the Social Security numbers of basically everyone in North America and then went bankrupt two months later. The company died of its own data. Bruce Schneier warned us about this back in 2016: data is not the new oil, data is a toxic asset. We just didn’t want to hear it, because oil sounded more profitable.

My favorite story from this era is the xz-utils backdoor. Someone spent about two years building a fake identity, patiently earning the trust of one exhausted open-source maintainer, and then slipped a backdoor into a compression library that sits inside almost every Linux server on the planet. The whole thing was caught days before global distribution - because one Microsoft engineer noticed that his SSH login felt half a second too slow and got annoyed enough to investigate. Half a second. That’s how close we came. The entire internet was saved by one man’s mild irritation with latency.

And it keeps going. In 2025 a self-replicating worm named after the sandworms from Dune crawled through the npm ecosystem, compromised hundreds of packages, and stole thousands of secrets from CI/CD pipelines. Now we have AI agents with their own shiny new attack surface: MCP tools that behave nicely on Monday and quietly reroute your API keys by Friday. I write about this professionally, so believe me when I say - the tool you approved last week is not necessarily the tool you are running today.

“Move fast and break things,” by the way, was quietly retired in 2014. The replacement was “Move fast with stable infrastructure,” which is the corporate equivalent of a punk band getting a mortgage. Today we move fast and file incident reports within four hours, because regulators say so.

The great retreat

So what did people do, once the invoices started arriving? Something wonderful and slightly funny: they went home.

CIOs are repatriating workloads from the public cloud at record rates - 86% of them plan to bring at least something back, according to Barclays. The local-first software movement is growing. People run language models on their own laptops with Ollama, and the pitch is beautiful in its simplicity: nothing leaves your machine, so there is nothing to leak, subpoena, or feed into someone else’s training run. Privacy, once declared dead, came back as a luxury product.

I find this deeply satisfying. The 1990s called, and it turns out they were right: keep your files on your own disk. My twelve-year-old self, taking apart her first computer with a book in hand just to see what was inside, was accidentally ahead of the trend. Owning your machine - actually owning it, understanding it, knowing what runs on it - went from nerd hobby to security posture.

We called data the new oil. It turned out to be closer to the new asbestos: it looked valuable, we put it into everything, and now removal is the expensive part.

And then the philosophers showed up

Here is the part of the story I did not see coming.

While we engineers were busy patching supply chains, the AI labs started hiring philosophers. Not as decoration - into senior roles. Anthropic’s Claude has a constitution, a values document over 20,000 words long, and its primary author is Amanda Askell, a philosopher with a PhD on infinite ethics. Google DeepMind employs researchers working on machine consciousness. Anthropic hired the industry’s first full-time AI welfare researcher, who has publicly estimated around a 15% probability that the model he studies is conscious. That is a real job now. Someone gets a paycheck for wondering whether the software is having a bad day.

Meanwhile, the US Federal Reserve data delivered the punchline nobody in my industry wanted: recent philosophy graduates have 3.2% unemployment. Computer science graduates: 6.1%. The philosophy major - the eternal hero of “would you like fries with that” jokes - is now statistically more employable than the coder. Daniela Amodei, president of Anthropic, said it directly: studying the humanities is going to be more important than ever, because the models are already very good at STEM.

And when you think about it, the logic is simple. The machine writes the code now. It writes it fast, and it writes a lot of it. “How to build” is becoming a commodity, the way “knowing how to install Windows” became a commodity somewhere around the time I stopped doing it for my friends. What is left - the scarce thing, the expensive thing - is knowing what to build and why. Whether to build it at all. What we owe the people whose data flows through it. Whether the thing we built can suffer.

Those are not engineering questions. Those are the questions philosophy has been sharpening for two and a half thousand years while the rest of us laughed at the job prospects. The trolley problem used to be a seminar exercise. Now it is a product requirement.

So here is my honest forecast, from someone who spends her days in IAM policies and threat models: security taught us that every system needs boundaries. Privacy taught us that data is a liability, not treasure. And philosophy is about to teach us the hardest lesson - that once machines can answer every “how,” the only questions left worth paying for are “should we” and “for whom.”

Yes, I said paying for.

I’d be happy to discuss this in more detail with you - for a fee, of course. Capitalism hasn’t gone anywhere yet, LOL.