Hey buddy,

Man, Friday on Hacker News had some wild stuff, lemme give you the quick rundown.

Grok on Musk's X Posts

First off, this guy used Grok, you know, Elon's AI, to search X specifically for Elon's own posts about Israel, Palestine, Hamas, and Gaza. Like, digging into what Musk himself was saying on his own platform about that whole situation. People in the comments were immediately diving into the politics, naturally, but also talking about how these AIs handle sensitive topics and whether they have "guardrails" like binding parameters in SQL, which was a kinda cool technical analogy someone made. Seems like the AI stuff is getting used for some intense analysis now. Check it out: https://simonwillison.net/2025/Jul/11/grok-musk/

OpenAI Deal Drama

Big tech news: OpenAI's deal to buy this company called Windsurf is off. And get this, Windsurf's CEO is now going to Google instead. Drama! Someone in the comments mentioned this tool called Augment that they use and think is way better than Cursor, which is an interesting alternative if you're into AI coding helpers. Another person compared this AI investment frenzy to the crypto bubble, which... yeah, feels about right sometimes. Read the story here: https://www.theverge.com/openai/705999/google-windsurf-ceo-openai

UK Post Office Scandal Gets Worse

Remember that terrible UK Post Office scandal where they wrongly prosecuted sub-postmasters because of faulty software? A new report came out saying at least 13 people died by suicide because of it. Just awful. The comments were talking about accountability – who's *really* responsible, the CEO or the board? And someone had a pretty cynical take on the whole "deep state" idea, saying it's not some hidden group, it's just... the state doing stuff like any other big organization. Heavy stuff. Link: https://www.nytimes.com/2025/07/10/world/europe/uk-post-office-scandal-report.html

Public LLM from Swiss Universities

Some universities in Switzerland, ETH Zurich and EPFL, are planning to release a big language model built on public infrastructure. Like, an LLM for the public good. The comments were a bit skeptical though, asking what the point is when there are already open-weight models out there. Someone else brought up the challenge of training these things, like how most current tokenization is really geared towards English. Interesting to see universities getting into the big model game though. Here's the announcement: https://ethz.ch/en/news-and-events/eth-news/news/2025/07/a-language-model-built-for-the-public-good.html

Japan's Overtourism Problem

An interesting article about overtourism in Japan and how it's actually hurting small businesses, not just annoying locals. The comments had a good debate going – some people defending tourism, others agreeing it's overwhelming and changes the vibe. Someone suggested Kanazawa as a less crowded alternative to Kyoto if you want that old-world feel. It really got people talking about the balance between tourism dollars and preserving local culture. Check it out: https://craigmod.com/ridgeline/210/

Click-to-Cancel Fight

Okay, this one's annoying but important. The federal "click-to-cancel" rule, which would make it easy to cancel subscriptions online, got blocked by a court. BUT, Pennsylvania's House just passed their *own* version of the bill. So now it's this state-by-state fight. People in the comments were complaining about dark patterns and how ridiculous it is that this even needs laws. Someone pointed out that implementing those annoying dark patterns probably costs companies *more* than just letting people cancel easily. Makes you wonder. Here are the links: PA Bill and Federal Block

McDonald's Password Fail

Dude, this is wild. Chats for 64 million McDonald's job applicants were exposed because they used the password '123456'. Seriously? '123456'? The comments were just shaking their heads, talking about how this is a classic "defense in depth" failure. Someone quoted the company's supposed security page saying "We worry about security, so you don't have to" which is just *chef's kiss* level irony after this. Also, people were facepalming that they just used sequential numeric IDs for the chats instead of something random like a UUID. Absolute mess. Read about the fail: https://www.bleepingcomputer.com/news/security/123456-password-exposed-chats-for-64-million-mcdonalds-job-applicants/

Alright, that's the main stuff. Pretty interesting mix today, huh? Catch you later!