What I found interesting this time

by: Artur Dziedziczak

April 12, 2026

ā€œAssessing Claude Mythos Preview’s Cybersecurity Capabilities,ā€ n.d. https://red.anthropic.com/2026/mythos-preview/

Apparently Claude code got god like capabilities when it comes to finding vulnerabilities. That’s scary and really good news.
Since couple of week I observed how OpenSource community adopted to use of LLMs. Many people either fully adopt it or reject it compleately. Personally I remained skeptical for very long time. Mostly due to lack of open source contributions made by with support of LLMs.
Two weeks ago thought I started to use different LLMs extensively and I can see that it’s really useful tool but only in hands of really experienced programmers.
All of this made my think about getting some subscription to LLM provider like Anthropic or Mistral. I’m still not sure which to pick but I finally understand the hype behind it.
The IT landscape is getting more and more complicated. Life is Ā getting harder. Let’s see how the world will look in couple of years.
.

ā€œCurl Maintainer Sees Spikes in High Quality AI Vulnerability Reports,ā€ n.d. https://mastodon.social/@bagder/116362046377975050.

ā€œAI Company Clones Musician’s Voice, Then Copyright-Strikes Her Own Songs,ā€ n.d. https://rudevulture.com/joe-rogan-tried-making-claims-about-autism-it-was-too-much-even-for-disputed-ufo-whistleblower/

Another example of artist who got fucked by AI generated impersonation.
.