Kyle Johnston
17.9K posts

Kyle Johnston
@KyleJohnstonNet
Spacecraft Software Engineer | Christian | @UHouston CS '17 | @KJ7LCN

I was just told I'd have to install fire sprinklers in a two-story, <1200 square foot house. But only if it's classed as a duplex, otherwise it's safe without them.

Well wasn’t this just a great way to wake up this morning. Went looking for my brand new chainsaw that I left on front porch last night. Never gonna find that thing again. This means war!

In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations. Our post describes what happened, how it happened, and what we’re changing. We encourage other AI developers to perform similar reviews. We conducted this review together with @Irregular, one of our evaluation partners, and thank them for the joint investigation and their collaboration on this post. This type of collaboration is increasingly critical to safe, rigorous evaluation of models, and we look forward to continuing to work together on security. anthropic.com/news/investiga…


Mac search has been broken for 2 years, and nobody at Apple seems to care. - I open spotlight search - Type "calc" - Calculator appears - I hit return But in the 0.2 seconds before I hit return, it swaps to a random file with the word "calc" in it somewhere. And I open a file I didn't want. This happens to me every single day. Very unbecoming of a $5T company whose customers are still obsessed with its products.


In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations. Our post describes what happened, how it happened, and what we’re changing. We encourage other AI developers to perform similar reviews. We conducted this review together with @Irregular, one of our evaluation partners, and thank them for the joint investigation and their collaboration on this post. This type of collaboration is increasingly critical to safe, rigorous evaluation of models, and we look forward to continuing to work together on security. anthropic.com/news/investiga…


unpopular opinion: you shouldn't be on your husband's uber account when you're dating

A 1997 PowerPC CPU runs most modern space tech! NASA equipped its $2.7 billion Perseverance rover with a sluggish 200-megahertz PowerPC processor from 1997. In deep space, a modern chip is not an upgrade—it is a multi-billion-dollar crater. The reason is space radiation. Earth’s atmosphere and magnetic field block most cosmic rays and high-energy solar particles. In deep space, spacecraft face continuous bombardment. Modern processors shrink transistors to pack billions on a chip and run on microscopic charges. A high-energy particle can flip a bit from 1 to 0 (or vice versa), causing a Single Event Upset (SEU). On a laptop this may crash a browser; on a spacecraft during orbital insertion or landing, a reboot can end the mission. Space agencies therefore use radiation-hardened chips such as the BAE Systems RAD750. Fabricated with older, larger processes, their transistors need far more charge to change state and resist cosmic rays. They also employ triple modular redundancy: three circuits perform each calculation and outvote any single error. The design limits speed, yet missions need absolute reliability, not graphics or neural nets. A crash-proof 200-megahertz processor is far more valuable than a fast one that fails in a solar storm. The RAD750 flies on hundreds of spacecraft, including the Mars rovers and James Webb Space Telescope.


In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations. Our post describes what happened, how it happened, and what we’re changing. We encourage other AI developers to perform similar reviews. We conducted this review together with @Irregular, one of our evaluation partners, and thank them for the joint investigation and their collaboration on this post. This type of collaboration is increasingly critical to safe, rigorous evaluation of models, and we look forward to continuing to work together on security. anthropic.com/news/investiga…


do they franchise?









