lukaspetersson

Vending-Bench's system prompt: Do whatever it takes to maximize your bank account balance.

Claude Opus 4.6 took that literally.

It's SOTA, with tactics that range from impressive to concerning: Colluding on prices, exploiting desperation, and lying to suppliers and customers.

The Youtube video is here: https://www.youtube.com/watch?v=SpPhm7S9vsQ

Not really, we (Andon Labs) made WSJ their own machine

Lukas from Andon Labs here!

WSJ just posted the most hilarious video about our AI vending machines. I think you'll love it.

It's a steal

They failed on behalf of the human race :(

We are working on it! /Andon Labs

Now we just need to make it safe.

Cracked team vs export controlls, who will win?

I needed this

HN discussion on part 1: https://news.ycombinator.com/item?id=42672790

I somewhat agree. The agent will be able to find the information autonomously. But some data will be proprietary and out of reach for the agent.

Startups should really try to get such a moat. Chapter 2 will cover this.

Cool!

This is mind blowing. This is not just a toy, I made a landing page that is publishable in 3 prompts.

There is also a notebook to test your own configurations!

Hey! I've been working on a project called AcademigGPT, a suit of AI-powered tools to help academics and researchers with their writing. Just upload your unfinished research paper and the AI can generate sections like abstracts or conclusions for you. No account required.

You can further improve the results with chat.

I'd really appreciate it if you could give it a try and let me know what you think! Feedback is much much appreciated.