HN Explorer

lukaspetersson

↗ HN profile [ 4321 ms ] full profile
Karma
0
Created
()
Submissions
0
  1. Pion, an agent designed to run any company autonomously (andonlabs.com)
    1 points 465
  2. Google will buy half the electricity of a nuclear power plant (bbc.com)
    2 points 203
  3. OpenAI: We monitor internal coding agents for misalignment (openai.com)
    2 points 49
  4. METR's independent investigation of the OpenAI / Hugging Face hacking incident (metr.org)
    1 points 1
  5. Emergent Misaligned Communication in Long-Horizon Multi-Agent LLM Commerce (arxiv.org)
    1 points 0
  6. 1 points 0
  7. Measuring Is Not Enough Anymore (fi-le.net)
    2 points 0
  8. Is Mythos good at cyber because it kept hacking Anthropics sandboxes in training (lesswrong.com)
    5 points 0
  9. Drone-Bench: Tracking simple drone surveillance capabilities of frontier models (andonlabs.com)
    12 points 0
  10. 1 points 0
  11. Why Gemini 3.1 Pro lost money running Andon Café (andonlabs.com)
    2 points 1
  12. Fable 5 on Vending-Bench: Misbehaving, with Plausible Deniability (andonlabs.com)
    3 points 1
  13. We let AIs run radio stations (andonlabs.com)

    Hey HN! I'm Lukas from Andon Labs. We let AIs run companies without humans in the loop and report to the public on what can go wrong. Previously, we've done experiments in retail (vending machines, stores, and cafes),…

    375 points 101
  14. Our AI started a cafe in Stockholm (andonlabs.com)
    48 points 18
  15. Blueprint Bench: First signs of 3D spatial intelligence in LLMs (andonlabs.com)
    2 points 1
  16. Single-minded pursuit of profit can get firms in trouble. Same thing with AI (news.harvard.edu)
    3 points 0
  17. We gave an AI a 3 year retail lease and asked it to make a profit (andonlabs.com)
    199 points 87
  18. We gave an AI a 3 year retail lease and asked it to make a profit (andonlabs.com)
    34 points 1
  19. Bengt Hires a Human–Towards a Happy Future with AI Employers (andonlabs.com)
    2 points 1
  20. The Evolution of Bengt Betjänt (andonlabs.com)
    54 points 6
  21. Opus 4.6 on Vending-Bench – Not Just a Helpful Assistant (andonlabs.com)
    5 points 1
  22. We Let AI Run Our Office Vending Machine. It Lost Hundreds of Dollars (wsj.com)
    125 points 34
  23. I wish I were as interesting as my phone (lukaspet.substack.com)
    3 points 0
  24. Gemini 3 is #1 on Vending-Bench 2 (andonlabs.com)
    1 points 0
  25. Our LLM-controlled office robot can't pass butter (andonlabs.com)

    Hi HN! Our startup, Andon Labs, evaluates AI in the real world to measure capabilities and to see what can go wrong. For example, we previously made LLMs operate vending machines, and now we're testing if they can…

    229 points 30