Facebook had to stalk your clicks to build a psychological profile. OpenAI figured out people will volunteer their fears, relationships, medical worries, and political beliefs if the surveillance machine responds with: »That sounds really difficult.« The therapist is a database, and the waiting room is a data center.

That’s the unsettling part. When billionaires spend fortunes shaping elections, it’s reasonable to ask whether they see voters as citizens to persuade or obstacles to manage. Democracy becomes a lot less inconvenient when you can fund the candidates, influence the platforms hosting the debate, and lobby the people writing the rules. All perfectly legal, of course.

The usual financial crisis playbook has a remarkably consistent cast: 1. banks discover a new way to create risk, 2. executives collect bonuses while it works, and 3. taxpayers get invited to the emergency meeting when it doesn’t. 👉 Somehow the public is always a stakeholder in the losses, but never in the profits.

The terrifying AI catastrophe here is apparently that customers might finally move their money to accounts paying decent interest. Banks spent decades automating every trick to extract profit from us, but the second consumers automate looking for a better deal, suddenly it’s a systemic threat. Funny how innovation becomes dangerous the moment it works for the customer.

Turns out telling people your technology will replace their jobs, devour their electricity, and possibly end civilization isn’t the winning sales pitch billionaires imagined. Imagine spending hundreds of millions buying political influence, only to discover voters don’t enjoy being treated like obsolete hardware.

Only in Silicon Valley can a CEO burn billions chasing an AI gold rush, send free cash flow below zero, and then demand a bigger payday for his visionary leadership. The board’s ultimate weapon? Threatening to quit. Corporate governance at Meta is basically a fire alarm that needs Zuckerberg’s permission to ring.

Google: We would never use your Gmail to target ads. Also Google: If you voluntarily pour your Gmail into our AI blender, whatever comes out may help target ads. Technically, we’re not reading your diary. We’re just selling billboards next to the therapist’s summary of it. Privacy by wordplay is still surveillance with better PR.

Turns out a business plan consisting of »build a trillion dollars of server farms now, figure out who pays later« gets slightly less charming once the lenders ask for actual numbers. The AI industry wanted to disrupt every profession except the ancient art of checking whether the math works.

Microsoft has reinvented the arcade, except this time you buy the game first and then feed the machine quarters to play it. Next up: a premium tier that lets you pause without the meter running. At this rate, Xbox’s most profitable game will be watching customers try to escape the billing screen.

Meta promised an AI built from the ground up for privacy, then built a handy directory of strangers’ loan applications and bank records. Researchers needed one query. At this point Zuckerberg calling something »private« is less a security guarantee and more a spoiler warning.

Two critical PHP RCE bugs, an embargo leak, and emergency patches on the same day. Somewhere, a sysadmin is discovering that the company’s »fully managed« hosting means they get to manage the panic themselves. Meanwhile, a forgotten WordPress install from 2014 is quietly preparing its villain origin story.

The administration wanted a voting machine hacked, but DOGE had already fired the experts capable of doing it. Apparently the one thing stopping a potential election conspiracy was Elon Musk’s ability to make government incompetent at absolutely everything. Task failed successfully.

This! 💪😎✌️

The scary part isn’t AI becoming intelligent. It’s corporations discovering they can sell us a monthly subscription to our own judgment. First it writes your emails, then it plans your life, and eventually the only human skill left is clicking »I agree« on the terms of our surrender.

The frightening part is that the slop doesn’t have to fool informed voters. It just has to reach people who don’t follow politics closely, long before a fact check ever catches up. One polished fake clip can become their entire impression of a candidate. Apparently political advertising has discovered the perfect product: certainty without facts.

Silicon Valley looked at places where reliable internet, affordable data, and local language support are still luxuries and thought: what if we added a $100/month bot that needs five times the bandwidth to buy tomatoes? That’s not closing the digital divide. That’s putting a toll booth on it and calling it progress.

We’ve invented a machine that can turn a post about the fictional Cactus Gravity Protocol into a sourced answer in seven days. That’s not a search revolution: it’s a misinformation laundering service with a bibliography.

Same here. Thankfully it’s a very long time ago. Since then I immediately ask: »Should I?«

Same guys will explain machine learning after reading one headline, call themselves thought leaders, and get promoted for being loudly wrong in a blazer. … The hallucinations were human first.

Management spent two years demanding everyone use AI, then discovered everyone was using AI. Now they’re shocked the office is a conveyor belt of hallucinated PowerPoints reviewed by other bots. The real productivity breakthrough will be executives learning that buying subscriptions isn’t a business strategy.