Nathan Hamiel

OpenAI’s write-up of the incident reads more like a marketing document promoting a feature than an incident summary, while the quote from HuggingFace sounds more like someone accepting an award than someone who just got hacked. It’s a bit surreal.

I swear Teams is being developed by sadists trying to implement extra-clickmaxxing.

Nils and I will be at the AI Summit at Black Hat USA. This time, bringing half presentation and half hands-on labs focused on defending developer AI environments. See you in Vegas! blackhat.com/us-26/summit...

People are obsessed with data, even when that data doesn't lead to useful optimizations. They often don't consider how data can be detrimantal to the activity they are collecting it for. In this article, I look at how this even extends to people who read. perilous.tech/reading-data...

Reorganizing some books this weekend and not sure how I ended up with two copies of Simulacra and Simulation.

And not a moment too soon. Companies were being torn apart by the vulnpocalypse! 🙄 Overhyping these aspects has turned out to be a fantastic foot gun. The features they want to protect against are the most useful features of the model.

This article has an error. What appears as addiction is actually dependence. These students’ brains are rewired by cognitive offloading, not by a chatbot's addictive design. Adding the friction back seems painful, so they fall back to using AI. futurism.com/artificial-i...

Rarely does a news story hit me so hard and go straight to my heart 😆

This is what AI discourse feels like in 2026 😆

"A spokesman for KPMG International said the firm takes the accuracy and integrity of its published content seriously.” 🤣 Content and services continue to degrade around us every day. Are you feeling the AGI yet?

Well, that was a pretty quick change. We didn't even need to wait 12 to 18 months🤣

Still true today.

AI won't make you irreplaceable. It might even do the opposite. In this post, I dissect the two most common AI phrases and provide some perspective on why you might not want to wave goodbye to your skills and become one of the faceless. #AI #Skills #Workforce perilous.tech/poor-advice-...

"To the slop architect, execution, skills, and experience are secondary, bowing at the pedestal of ideas. The fact is, most ideas are ill-thought-out, half-baked, or just plain fucking stupid." perilous.tech/the-architec...

I see we're doing this again.

This take from Ed Zitron needs some tweaking. LLMs impress the non-writers who want to write, the non-coders who want to code, the researchers who simply want to boost their publication count, and the lawyers who’d rather be drinking.

This may be a drop in the bucket for Google, bit it says a lot since it takes a lot for people switch. However, when a service continues to degrade with a slew of features nobody asked for, that's what happens. BTW, you still can't Google the word "disregard" in AI mode.

The flip side is that with the guardrails removed, the models become more useful. Much of the commentary about the dangers here is overly sensational. The way guardrails are added to models is more like wrapping paper than a true protection mechanism anyway.

This is a follow-up to my techno-communism post from last year. So many people talk about utopias or dystopias from the perspective of already being in them. What about the time leading up to them? This is what I call the dystopian lag. perilous.tech/the-ai-adjus...

Pull out irony gun, point directly at foot. He started his own investigation? Wasn't that what writing the book was? Seriously, he had one job! 🤦‍♂️ Incidents like these are a lot more damaging than people think.

Nope, the singularity isn't 4 years away, but the shitularity is certainly already here.

These predictions are always 12 to 18 months. That's because the timeframe is short enough to seem impactful, yet long enough for you to forget about the prediction after it doesn't happen.

Ah, yes, "innovation." Please capture what's on my screen, what I say, and what I point at all so I can... do what exactly? It's obvious what Google gets, but what do I get for all this surveillance? Other than, zomg! My Ctrl-c has AI!!!

Right on queue, the democratization of AI psychosis continues. Dawkins was hit by what I called local bias. He blew himself away with his experiment. perilous.tech/democratizin...

Imagine having to pay a token ransom to keep your solutions functional. This may become a reality in the very near future. AI cost volatility is a new risk that must be factored into the calculus of AI use and the development of AI-powered solutions. kudelskisecurity.com/modern-ciso-...

Agents don't just hallucinate or create errors. They do so with action. If your design isn't robust against these inevitable failure modes, you're gonna have a bad time.

Okay, who made these icons? 😆 When I called out that the icons shouldn't be used, I was asked why. Upon realizing I'd have to explain, I said, "Forget it. They look great."

Oh, great! Happy Friday, I guess.

This article examines the democratization of AI psychosis and the capture of smart people by AI hype. A fascinating yet terrifying trend. We need to recognize when our perceptions of reality are warped, and it’s not as easy as it sounds. perilous.tech/democratizin...

What's happening with all of these price hikes and capability reductions in AI products? Influencers and "visionaries" have been telling us to plan for the price of intelligence to be zero! Apparently, subsidizing tokens while hemorrhaging money isn't a great business strategy.

Since I discuss risks and hype so much, there seems to be some confusion around my perspectives on AI. I’ve tried to document some of my thoughts here. I’m not a skeptic. I’m a critic. perilous.tech/my-perspecti...

Outrageous claims are made regarding AI and abundance. Claims so absurd that they have to be true, right? Well, other than the math not making sense, there are lessons we can learn about abundance from Napster. Because when things become free, they are worth nothing. perilous.tech/what-napster...

Recently, there have been plenty of slot machine analogies applied to AI, but Jacques Ellul made this association back in the 1950s. He observed how humans participate less and less in technological creation and are reduced to mere catalysts.

My cat says no conference calls today!

I may still write about this zero-introspection situation, but this is an excellent article. One omission is that many people attribute to skill what should be attributed to chance. "You’re a 15 second sliding context window." 😆 www.theverge.com/tldr/897566/...

People are losing their jobs to AI, not because of its capabilities, but because of the mere idea of AI. Companies that AI-wash layoffs see short-term gains, but these may translate into long-term losses. Let the great AI washing begin. perilous.tech/the-great-ai...

Time is closing in. The Black Hat USA CFP closes this Friday. For the AI track, we are looking for offensive, defensive, and applied topics. Let's see what you got! Let me know if you have any questions. blackhat.com/call-for-pap...

Does someone need to take Zuck's phone away from him? No nanna, that's not a real Nigerian prince!😆 gizmodo.com/mark-zuckerb...

The next few years will require vigilance and the ability to envision trade-offs even when no evidence of trade-offs is apparent. These are essential skills in a world that prioritizes dehumanization. This starts with not confusing innovation with progress. perilous.tech/confusing-in...

We need to get much better at envisioning tradeoffs. A symbiosis with AI would mean that we would never know if a thought or memory we have is truly our own. It’s the end of private thoughts and the beginning of a whole new world of manipulation and unintended consequences.

I first met FX in the early 2000s. We had so many laughs, so many memories. Hell, during just one notorious hacker trip in 2009, there were enough memories to last a lifetime. He will be missed.

The biggest hot take of the past few weeks is that software is dead. But is it really? Seems there are some fundamental realities not being considered. Regardless of success, software vulnerabilities will be absolutely everywhere. Welcome to the new reality. perilous.tech/the-death-of...

This Clinejection write-up is great, and I learned some things about GitHub actions caching, too. We experienced the same during our research for our Black Hat USA 2025 talk on attacking AI-powered developer productivity tools. adnanthekhan.com/posts/clinej...