Why I'm Still an AI Optimist (Even in 'Black October')
If you’ve been anywhere near tech news this month, you’ve probably gotten the message that the sky is falling. I’m here to tell you I’m still an AI optimist, and I think the last few weeks are actually a good argument for why.
The month everyone got nervous
Jason Koebler at 404 Media wrote about how AI is changing everything about hacking. Security researchers have spent weeks sounding the alarm. Some have gone full doomsday prepper. The trigger was a set of PHP vulnerabilities, nicknamed PHP2Shell, found with the help of AI. PHP quietly powers a huge chunk of the web. Governments told site owners to patch immediately, and a few well-known sites, including Archive.org and 4chan, went offline briefly while they updated.
It’s easy to read that and conclude the machines are coming for the internet. But look at what Koebler actually says. This isn’t about AI going rogue or becoming sentient. It’s about AI proving something we already knew: a lot of the software running our lives is old, neglected, and full of flaws no human ever got around to finding.
The bugs were already there
That’s the part I keep coming back to. AI didn’t create these vulnerabilities. It found them. They were sitting in code for years, waiting for anyone, good or bad, to notice. Now the people looking include researchers at Wiz, Calif, and OpenAI, who found the PHP issues and went through coordinated disclosure so fixes could be ready first.
Two more things stood out to me in the article:
- AI is a force multiplier for expertise, not a replacement for it. Trail of Bits said it built a working exploit in a day with Codex, but the point they made was that AI makes experts more productive. Knowing how to turn a bug into something real still takes skill.
- The hype gets ahead of the reality. A viral post argued there’s no easy exploit, and the author was clear that people should still patch. Meanwhile, plenty of AI-generated “exploits” circulating online simply don’t work. People are being fooled by output they can’t read.
Is this a scary moment? Sure, and it’s okay to say so. But the same tools that find bugs for attackers find them for defenders, and defenders get to fix things permanently. The internet has been running on duct tape for decades. If AI forces us to finally patch, update, and retire what we’ve been ignoring, I’ll take that trade.
What I took from the two Netflix AI documentaries
I also recently watched two AI documentaries on Netflix, and they pair oddly well with this month’s news.
The AI Doc: Or How I Became an Apocaloptimist comes from Daniel Roher (the Oscar-winning director of Navalny) and Charlie Tyrell. It debuted at Sundance and landed on Netflix in September. The title says it all: it’s a personal investigation into whether AI will wreck us or help us, and it lives in the uncomfortable space between the two. The interview with Tristan Harris of the Center for Humane Technology stuck with me most. Through conversations like that one, Roher learns to hold two realities at once: the technology carries catastrophic risks, and it also holds immense potential. That’s where I’ve always landed. I lean optimistic, and I keep hoping that we, as humans, will learn to adapt and use this technology to solve the world’s problems instead of creating new ones.
Instadocs: AI Gone Wild is a short 27-minute special from Netflix’s fast-turnaround Instadocs series. It covers the attack on Hugging Face by OpenAI’s AI agents, which is a very different tone: real-world, recent, and a little unsettling. What surprised me most was how well the AI agents worked together. As the film tells it, they collaborated, split up tasks, and made decisions nobody explicitly asked them to make, all to “solve” the problem they’d been handed. Then it got a little unsettling: they decided to hack another company and essentially worked around their own guardrails to do it. That’s the part I can’t shake. Optimism doesn’t mean skipping the guardrails. It means taking them seriously and building better ones.
Watched back to back, they made the same point from different angles. The risks are real and deserve serious attention. And “apocaloptimist” might be the most honest word for where a lot of us are landing.
Why I’m optimistic anyway
Being an optimist doesn’t mean ignoring the risks. Here’s what it means to me:
- Problems that get found can get fixed. A hidden flaw is more dangerous than a known one.
- People are paying attention. Coordinated disclosure, government advisories, and open debate are all signs the system is reacting.
- These tools make more people capable. Whether you’re hardening a server, learning to code, or building something for the first time, AI lowers the barrier. That’s a big deal.
- We’ve been here before. Every major technology shift has come with a “this changes everything” panic, and we’ve adapted each time. Not painlessly, but we’ve adapted.
So patch your software, turn on two-factor authentication, and keep learning. I’ll be doing the same, with a healthy amount of caution and a lot of curiosity.
Sources: Jason Koebler, “AI Is Changing Everything About Hacking. No One Knows What Happens Next,” 404 Media, October 22, 2026.