You gotta hear about some of the stuff from Hacker News yesterday, Tuesday. It was pretty wild, man. Grab a coffee, here’s the rundown:

Ghostty 1.4 Dropped – Terminal Talk

First off, you know that terminal app, Ghostty? They just dropped version 1.4. People are saying it's super fast and lighter than stuff like iTerm, which is cool.

But get this, some folks are still seeing occasional rendering glitches, and it's not as good at spotting URLs as other terminals. And there's this whole debate about something called OSC8 for security – apparently, it could open up new holes, which is a bit gnarly for something you use for sensitive stuff. Oh, and a bunch of people are still swearing by WezTerm, especially for its keyboard shortcuts.

US Government Hiding How It Spies on Americans

Then, big one: Wired reported that the US government is apparently hiding a secret court ruling about how it spies on us, you know, Americans. Super sketch.

The comments were wild. Someone who actually attended parts of the trial chimed in, saying it's not about how they spy, but whether it's even legal in the first place. Like, they're not hiding tech, they're hiding a legal judgment. Crazy, right?

DeepMind's Genie 4 – World Simulation AI

DeepMind dropped 'Genie 4,' their new 'world simulation' AI. Sounds fancy, like it could make games or something.

But man, the comments were pretty unimpressed. A lot of people were like, "This isn't really a world model, it's just really good at faking it." They're saying it's still got a fundamental gap, just better at masking it with more compute. So, not quite Skynet yet, I guess.

AI Promised Less Work, But I'm Working More

Remember all the hype about AI making us work less? Well, there's this article titled 'AI promised less work. I've never worked more.' Super relatable, right?

The comments were interesting. Some people are saying AI just lets them cram more work into the same day, not actually work fewer hours. Others are talking about being 'centaurs' – where you delegate small tasks to AI but keep control of the big picture, otherwise you burn out.

Anthropic Trained Claude on Pirated Books?!

Okay, this one's a bit spicy: Anthropic, the folks behind Claude, apparently trained their AI on illegally obtained books from Stanford professors. Like, pirated stuff!

People in the comments were bringing up Aaron Swartz, saying it's wild that a big company like Anthropic can do something similar with less backlash, probably 'cause they have the money and resources. And some folks were even like, "So what? It's just a book, it's meant to be read by everyone, even computers." Wild take, I know.

Python's Experimental JIT to Be Removed

For the dev side, Python's experimental JIT compiler is getting the axe. Apparently, it's just too hard to retrofit into Python's architecture.

The comments had a lot of 'I told you so' vibes, with people saying Python's just not built for super-fast optimization, and that folks doing heavy-duty stuff are already using tools like Numba anyway. Some were wishing it was more like Julia, where everything compiles on first launch.

Police Using AI for 'Suspicious' Driver Behavior

Last one, a bit dystopian: Police are apparently using AI to pull over drivers for 'suspicious' behavior. Like, what even is suspicious behavior to an AI, you know?

People were debating if this is like Waymo cars just stopping, or actual AI-driven profiling. Definitely a bit unsettling.

Anyway, that's the gist! Catch you later, man.

Later!