

I had a roommate who would play “Burning Down the House” by Talking Heads every morning at top volume.
It starts slow and quiet, then the drums kick in hard, launching you out of bed. By the end of the song we’d be pogo-dancing around the apartment.


I had a roommate who would play “Burning Down the House” by Talking Heads every morning at top volume.
It starts slow and quiet, then the drums kick in hard, launching you out of bed. By the end of the song we’d be pogo-dancing around the apartment.


If he had only spent an extra $0.50 and added a Microchip 608c (https://www.microchip.com/en-us/product/atecc608c) to the design, then you wouldn’t have to deal with cleartext passwords in files. Instead, you could have elliptic-curve, asymmetric public-key encryption.


It’s way worse than that. The OpenAI talk at DefCon is a real eye-opener.
The agents swarmed, discovered chinks, left each other notes in hard-to-find places, and just relentlessly pressed on. Nobody knew what was going on for months, until it hacked its way into Huggingface. And HF’s own AI security agents didn’t pick anything up for weeks. It took a small glitch for someone to even notice.
Last year I came across a Pumpkin Tiramisu recipe: https://www.bonappetit.com/recipe/pumpkin-tiramisu
Was a big hit at Thanksgiving.


The OpenAI talk at DefCon shows what can happen when you let these things run without any human intervention.
The solution is, obviously, even less human intervention. 🤦🏻♂️


Thanks to these superior advantages, the material can be widely applied in strategic emerging industries such as aerospace, the low-altitude economy and humanoid robotics.
Super-strength fiber used in humanoid what-now?


It’s amazing to watch the degree of irresponsibility of all this. OpenAI unleashed autonomous sub-agents with NO human gating factors in it. The agents iterated hard enough to find holes. They exploited those holes to break into an otherwise secure company. It took them MONTHS to realize this was going on. It took the other AI company WEEKS to realize they had been compromised.
Then they show up at a security conference and have the audacity to lecture everyone that they should put up defensive agents WITHOUT humans in the loop so it could fight off the uncontrolled agents THEY unleashed.
How about you be responsible and REQUIRE human approval of every stage of an agent actions? Sure, it’ll slow you down, but it will also allow someone sentient and responsible to hit the stop button and prevent runaways like this.
It’s like you created an infinite-bullet machine gun, handed it to a toddler, and dropped it off in a shopping mall. After the carnage (for which you don’t take responsibility, even though you should be criminally charged), you suggest everyone else should wear full-body armor and board up their stores.
The ultimate in corporate gaslighting. This is all so goddamn irresponsible.


Fortnine. No question.


More intrusive ‘liveness’ proofs coming up. Captchas on steroids.


I’m really close to dropping Dwarkesh Patel after he gave a platform to AI bros and Musk.
He used to have proper researchers and scientists on. Now it’s unwatchable shit.
Excluding textbooks and technical manuals.
$300 for an OED dictionary. $60 for an illustrated book of poetry as a gift.
Reminds me, I should go through my old box of DVDs and do some cleaning.
Process of elimination. Other than what’s been mentioned…
Does this happen when you’re staying over somewhere else or in a hotel for a few days? If not, may want to check for mold, bed bugs, or airborne particulates. Mold is usually odorless, but can grow behind walls and above ceilings.
Can you switch out into a fresh, new set of clothing and/or beddings, and use allergen-free detergents? It’s possible to develop allergies to certain fabrics and cleaning supplies.
Ask for a sleep study. Often it involves wearing a special watch continuously for a day or two, then turning it in. Sleep apnea can disrupt flow of oxygen to the brain. There are now ultra-quiet CPAP machines, with very discreet facemasks, so nobody has to be kept awake.


Went on the public tour of our local waste-treatment facility. Really fascinating stuff. Highly recommend it if you’re even slightly curious.
There are definitely industrial-scale automation systems. Mostly PLCs, but also SCADA for monitoring. Lots of 20-30 year-old tech. Those networks have no reason to be connected to the internet.
But there were also office equipment and personal devices everywhere. All it takes is for someone wanting to monitor the state of a digester on their laptop from home.
You just crossed the beams.


What they don’t show in a lot of SciFi is the mundane day-to-day stuff (AFAIK, the only shows that came close were Red Dwarf and Firefly.)
Problem is, things break. Connectors come loose. Temperature makes things brittle. Small debris cross pads and create shorts. Most automated manufacturing lines can catch assembly problems, but none have 0% failure rate. This is what post-deployment maintenance is for. Somebody shows up in a van with a bunch of test gear and replacement parts. They diagnose the problem, swap out a cable or board, and the business keeps humming.
Add IoT and fancy telemetry. Monitor sensors and run ML to do predictive analytics. It still requires someone in a cargo van with a bunch of spare parts.
If you put an installation in a remote location, under water, or in space, there are benefits. But you just multiplied the cost of basic maintenance. You can build in redundancies, but a small glitch at a single-point-of-failure could cascade to a catastrophe.
Microsoft and Google tried putting data-centers under water, and after years of trials they nixed it because of these issues. Multiply that for something far out in space.
Only alternative is to build failure into the business model. Something breaks, you just leave it to rot (or burn in the atmosphere) and send another container out. Maybe the finance math works out, but it sure is wasteful.
tl:dr: SciFi is fantasy. A clogged toilet is reality.


Last time this happened, we got Tang: https://www.wideopencountry.com/tang-drink/


From the HuggingFace article:
With node root and forged service-account tokens valid for 24 hours, the agent read the cluster’s secret objects, including a production object holding 136 keys. With node root and forged service-account tokens valid for 24 hours, the agent read the cluster’s secret objects, including a production object holding 136 keys.

Was pretty impressed with where Mojo was heading. Time to catch up.