Annoyed my work didn’t go down today

 

The Los Angeles Superior Court website was/is mostly down due to both 1) the Crowdstrike update, and 2) a separate ransomware attack.
Interesting in a clinical way-- one cannot access tentative rulings or the department calendars, but the Teams-based virtual appearance connection seems to be working. That means lawyers can appear in court, but not know when their cases are up or how the judge is inclined to rule.

And yes, I have a critcal game-ending hearing tomorrow (Monday) at 8:30 a.m., and the other side appears to have violated a prior court order.

 

184 Million Passwords Leaked for Google, Facebook, Instagram and More.
https://www.cnet.com/tech/services-a...your-accounts/

 

AI LLMs are really good at finding 0-day vulnerabilities. If you point Opus at a program and tell it to analyze a specific file in that program, it's good at finding uknown flaws in the file and writing exploits for it. Anthropic says that they've already found 500 high severity vlunerablities in a variety of programs. They've reported them and patches are coming out.

So now Anthropic is going to try and restrict people from doing this but Pandora's Box has been opened. This could be bad. Hopefully all these large companies start using this internally to find and fix their own software

 

AI models have found several major Linux kernel vulnerabilities in just the past week or so and have caused a major patching backlog to Mozilla firefox. I’m ambivalent about AI in general for most things, but this is definitely one of excellent use cases for it. Granted most of these vulnerabilities were never spotted by or exploited by humans (that we know of!) but it’s good to find them much faster with this technology, than with human code audits alone.