Man, you won't believe some of the stuff on Hacker News today, Wednesday, June 17, 2026. Had to call you up real quick.

New Version Control – Lore

First up, there's this new open-source version control system called Lore. It's supposed to be super scalable, which sounds like it's trying to tackle those big files and lots of changes, kinda like what game studios deal with. Some folks in the comments were saying how challenging it is for proprietary stuff, and how companies don't really care to manage it. Someone who's used Perforce for like 21 years in game dev mentioned they rarely even use the 'reconcile offline work' feature, so maybe Perforce already handles some of this pretty well. Another person pointed out the website doesn't really show typical use cases, which is a bit of a bummer if you're trying to figure out what it's for.

AI Marketing is a Turnoff

Then, get this, a study says sixty percent of US consumers are totally turned off by 'AI' in brand messages. People are just sick of hearing about it, I guess. One commenter had a good point though, saying nobody *wants* to talk to a human customer service agent either if the AI actually works. It's more about poorly used AI, not AI itself. But someone else said they just stop reading if something even *feels* like AI wrote it, especially after getting bad AI summaries from a law firm! Yikes.

New AI Model GLM-5.2

On the AI front, there's a new open-source model called GLM-5.2 that's apparently topping the charts on Artificial Analysis. Sounds pretty powerful. But, one developer who codes with AI daily actually said GLM-5.2 was 'mediocre with hand holding and super slow' compared to other models they use. So maybe benchmarks don't always tell the whole story for real-world use.

US Science in Chaos

Switching gears, there's a big article from Scientific American claiming 'U.S. science is in chaos,' saying the link between science and politics is broken. Sounds pretty serious. Lots of debate in the comments about whether you should just 'trust the experts' or always look at the best models that explain observations. Someone even brought up copyright law as a big problem.

Volkswagen Blocking GrapheneOS

Here's a weird one: Volkswagen apps are apparently blocking users who run GrapheneOS, which is a privacy-focused Android version. So if you're trying to keep your phone super secure and private, VW might not let you use their car apps. A commenter explained that companies often treat software like hardware production lines, and they're super worried about liability and lawsuits, which makes them wary of open-source stuff. Someone else said GrapheneOS is compatible with almost all Android apps unless they actively block it or have bugs caught by its protections.

Paying for Your Own Images Back

Another wild one: some service is charging users five bucks just to get their own images back after they cancel their subscription. Talk about holding your data hostage! Someone mentioned it's like other companies that penalize you for canceling, like those long-term subscriptions with big fees. And a person from the EU was happy they can just request all their data under GDPR, which is pretty cool.

OpenAI Losing Billions

And finally, big news for OpenAI – apparently, leaked financial docs show they're losing billions of dollars a year! Massive compute burn, too. Seems like building these huge AI models is incredibly expensive. A few people compared it to Amazon or Uber in their early days, which also bled money for years before becoming profitable. But another comment was like, 'there's no moat' because people can self-host decent LLMs on cheaper hardware now. Interesting times for AI companies, for sure.

Alright, gotta run, but thought you'd want to hear about that stuff! Talk soon!