Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. HackMIT 2026 (hackmit.org)
    —discuss
  2. The Collie and the Pug – HackMIT 2026 Keynote (3blue1brown.substack.com)
    —discuss
  3. Philosophy Bench (philosophybench.org)
    —discuss
  4. Claude's remake of "I'm upping my p(doom)" in 12 hours (x.com)
    1comments
  5. BigQuery FinOps Framework (alvin.ai)
    —discuss
  6. Unexplained objects in 1950s sky correlate with nuclear tests and UAP reports (nature.com)
    1comments
  7. SpaceX Pivots Away from Space (ft.com)
    —discuss
  8. AI for Creativity (twitter.com/jeffreykwndr)
    —discuss
  9. Astra compressed 600Mb of audio into 20Kb (twitter.com/matternjustus)
    —discuss
  10. A Mass. lesbian bar required KN-95 masks. When rules changed, hell broke loose (bostonglobe.com)
    1comments
  11. OpenAI $500 ProMax plan listed in API
    —discuss
  12. Musk – Official Trailer – Bleecker Street [video] (youtube.com)
    —discuss
  13. Steppy Fox Browser Game (steppyfox.com)
    —discuss
  14. Legal AI Harvey's margins went from +50% to -50% in 6 months on frontier costs (yahoo.com)
    2comments
  15. Show HN: IGViewer – Browse public Instagram profiles and stories anonymously (igrecent.com)
    —discuss
  16. Real-Time AI Simplification (A2, B1, B2) of BBC World News Live (voicefrom.ai)
    —discuss
  17. My First Language Front End with MLIR (whereisalan.dev)
    —discuss
  18. Dutch governments builds alternative for Microsoft based on NixOS (dawo.community)
    —discuss
  19. You Cry [video] (youtube.com)
    —discuss
  20. Your Language Model Is Already a Decision Model (github.com/ntlm1686)
    —discuss
  21. Dating a takeoff photo to the second using runway markings as a ruler (spuria.app)
    1comments
  22. Show HN: Can an AI agent bypass a post-quantum signed authorization policy? (github.com/max-russo-com)
    —discuss
  23. Coming to terms with the end of the Falcon 9 (thespacereview.com)
    —discuss
  24. OpenAI prepares new $500/month Pro Max plan for ChatGPT (testingcatalog.com)
    —discuss
  25. Show HN: Foveon Lab – Compare Sigma Photo Pro Presets on Foveon X3F Files (github.com/biopoetic)
    —discuss
  26. Ask HN: Migrating personal email domain from Gmail to iCloud?
    —discuss
  27. Deep Fission ($FISN): A Money Pit (iceberg-research.com)
    —discuss
  28. Show HN: Tasket++ Simulate your Windows actions automatically-free and OS (apps.microsoft.com)
    —discuss
  29. Tesla 'FSD' sped in 55% of Brussels 30 km/h zones ahead of EU vote (electrek.co)
    —discuss
  30. Driver Misconduct in Last-Mile Delivery (arxiv.org)
    —discuss

Tell HN: No company scaled revenue as fast as AI companies

2 pointsby 2h ago
2 comments
No company in recorded business history has scaled recurring revenue as fast as OpenAI and Anthropic. Forecasts show AI is advancing far faster than all human predictions. If you don't use paid frontier AI daily, your mental model is likely far behind where it currently is. Even for those of us who do use it daily, we are shocked by the massive intelligence gains every time we use it.

I am building a serious program using AI and I have yet to write a single line of code. AI has resolved 99.9% of bugs without me, and has come up with 95% of the features and roadmap. If AI costing a few hundred dollars can do this, what is going on inside the AI companies is beyond imagining.

We are not on a linear curve, we are on an exponential curve. One day soon, perhaps today, it will be so far further than human intelligence that we will never have another chance. Our only chance to survive may be right now today or tomorrow. Human institutions like Congress are not built to handle exponential curves like this. Please call, email, or visit in person every single member of Congress you can right away. If you are working in an AI company, pull the plug now.

The best experts in AI give us 10-50% chance that it will kill every man, woman, and child. We only get one chance to get this right.

2h agoHN ↗

How does AI go from being able to produce software with 95% of the features, to killing every man, woman and child?

People need to stop it with the ai doomerism.

Climate change will kill way more people than ai ever will, but you dont hear a pip about that from those claiming ai apocalypse.

1h agoHN ↗

Yet somehow, the U.S. Military still can't take a current satellite photo of their targets to confirm they're not about to murder 100+ innocent grade schoolers, and blame an overreliance on AI for the American atrocity instead.