New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Text Recognition techniques for premodern Italian and Devanāgarī manuscripts(uniqueatpenn.wordpress.com ↗)
    1comments
  2. A Super-Accurate Clock Using a Tiny Microcontroller(hackaday.com ↗)
    discuss
  3. The structure and lasting impact of the Roman road system(nature.com ↗)
    discuss
  4. Jev means structured output is interesting again(seangoedecke.com ↗)
    discuss
  5. Qwen3.8 Max – Cost per task higher than Astra on Artificial Analysis(artificialanalysis.ai ↗)
    1comments
  6. COBOL dev won .NET hackathon with help from AI(theregister.com ↗)
    discuss
  7. Valeriepieris Circle(wikipedia.org ↗)
    discuss
  8. Show HN: Point at anything in your app, say what you want, get a spec(dhamark.com ↗)
    discuss
  9. Show HN: LaunchPad – I built a Launchpad to replace macOS Apps
    discuss
  10. UK mathematician debunks myth around Parthenon's optical illusions(theguardian.com ↗)
    discuss
  11. The Gap Will Grow(backnotprop.substack.com ↗)
    discuss
  12. The secret formula for Apple's rounded corners (2023)(arun.is ↗)
    discuss
  13. Dense vs. Moe Models: Active Parameters, Throughput, and When to Choose Each(nvidia.com ↗)
    discuss
  14. Nightmare Eclipse revelaved his identity(twitter.com/msnightmare2000 ↗)
    discuss
  15. Secret Life of Circuits – Illustrated Guide to Electronic Circuit Design(nostarch.com ↗)
    discuss
  16. Sanders: "If AI is coming for workers, it should bring a 4-day workweek with it"(fortune.com ↗)
    1comments
  17. Ask HN: How do you test an app from a region you're not physically in?
    1comments
  18. Attribution-Based Control in AI Systems(attribution-based-control.ai ↗)
    discuss
  19. Why WordPress Is Still the #1 CMS in 2026(wpaudit.cc ↗)
    discuss
  20. Apple M5 Ultra Geekbench(geekbench.com ↗)
    discuss
  21. Roundtables: Could AI kill us all?(technologyreview.com ↗)
    discuss
  22. Ask HN: How do you get coding agents to clean up after changing direction?
    discuss
  23. DFOS is an open protocol for identities, signed records, delegated permissions(dfos.com ↗)
    discuss
  24. Klaus Programmieren(klausprogrammieren.com ↗)
    discuss
  25. Due to DMA, Siri AI delayed in EU for iOS 27 and iPadOS 27(apple.com ↗)
    discuss
  26. Mozill's Built-In VPN in Firefox
    discuss
  27. Bootstrap from Twitter(getbootstrap.com ↗)
    discuss
  28. GPT Image 2.5 Prompts(github.com/vulcaneon ↗)
    discuss
  29. How Streaks Work?(sheets.works ↗)
    discuss
  30. How meaning is fundamentally altered by sonic environments(danieldeboulay.com ↗)
    discuss

Sam Altman says world 'right to be afraid' but 'should trust' AI firms

3 pointsby 2h agobbc.com
2 comments
1h agoHN ↗

Trust is earned. Stealing all the content possible to train the models, then advertising that the model created from that stolen work will be used to eliminate everyone’s jobs, destroys any and all trust.

Earning back the trust of the public will take decades of consistent action that demonstrates trust. Asking for trust is worthless.

People still don’t trust Zuckerberg, starting with his, “they trust me… dumb f*cks” comment 20 years ago. To come back from something like that, a person has to go above and beyond, consistently, for decades. Zuckerberg didn’t do that. Altman isn’t doing it either, not that I can see.

It’s also not just about Altman. The entire industry needs to rebuild trust, and I don’t think any of them see trust as valuable. They want the trust, but they don’t want to put in the work or sacrifice to get there. For them, from where I’m sitting, this idea of trust is more of a strategic business advantage they want, not a deeply held set of principles or morals.