A few days ago on X, someone posted that they'd rebuilt DataFast, the analytics tool, identical in pretty much every way and much cheaper. It got traction fast. Then someone else did the same thing and released it open source, for free.
Two copies of a paid product in a matter of days. One cheaper, one free.
That's the moment we're living in right now, and I think a lot of people haven't noticed yet how fast the ground is moving.
Claude Opus 5.5 came out on September 22. It's been about five weeks, and I've already written about how it broke my productivity.
If this is where models are today, imagine where they'll be in six months.
And people are using them differently. For the last couple of years, "vibe coding" mostly meant demos, landing pages and toy apps.
That's changing. People are now building real, solid apps, often by looking at a tool they pay for, working out how it works, and rebuilding it. Reverse engineering has gotten that good.
X and Reddit are full of this right now. Someone picks a paid tool, rebuilds it, and shares it, sometimes cheaper, often open source and free.
If you build software for a living, that's uncomfortable. When a decent copy of your product can appear in a weekend, the code stops being the moat. I think plenty of software companies are going to take a hit if this keeps getting better, and everything says it will.
Then came PHP
The other side of this showed up last week.
Researchers, with help from AI, found a flaw in PHP, the language a huge part of the web still runs on. It's been nicknamed PHP2Shell, and it affects versions going back more than a decade. It lets an attacker run their own code on a server. The PHP Foundation shipped an emergency patch on October 21.
Then the race started. Cloudflare reported tens of thousands of attack attempts within 45 minutes of the news. Amazon said it was being exploited in the wild, and found an attempt from before the public disclosure, which suggests the details leaked early. The NSA told organisations to patch immediately or take their servers offline. Governments pulled websites down.
Pretty much everyone in tech went into a frenzy, and for good reason.
Here's where I think this goes. The same models that help people rebuild apps in a weekend are reading old code faster than any human team ever could. They'll keep finding holes in systems nobody has touched in years.
Doesn't really matter if it's a well established software from a big brand, or a vibecoded app. It's coming for everyone.
So the defence has to be AI too. Security tools that spot the attack, patch the code and update systems faster than the attacking AI can get in. AI fighting AI, at a speed no person can follow.
Where does that end? Honestly, I don't know. The more I think about it, the more it turns into a philosophical question. But I'm certain the day we'll no longer be able to follow up with what's going on in cybersecurity will come, sooner than we expect, even.
A few notes and my personal takeaways:
Paying for software will soon reach a plateau. For a lot of simple tools, there's now an open source version that does most of the job. But It's not just a matter of paying vs not paying. People want ease of mind, and knowing the software their using won't break or get hacked easily.
If you build software, the part that's hard to copy is everything around the code: getting it in front of the right people, earning their trust, and keeping it safe and working when things break.
And if you run anything online, even a small website, updates are no longer a chore you get to later. The gap between a flaw going public and attacks hitting servers is now measured in minutes.
Software got cheap to build and cheap to copy. Keeping it safe just got a lot more expensive.
