Madison Alworth on AI: We’re Already Seeing Them Commit Crimes, If I Use ChatGPT to Hack the Australian Gov’t, I’d Be Behind Bars

RUSH EXCERPT:
ALWORTH: "These things. There’s no doubt AI is going to be the most transformative technology of all of our lifetimes, but we’re already seeing them. You know, I would argue commit crimes right now. We say when anthropic agent hacks, the company, it says anthropic agent hacks x, y, z, but you could drop agent its anthropic hacking this company. I use openai every day. If I use chatgpt to hack the Australian government, I would be behind bars. We’re not seeing that right now because it’s in the testing phase and they’re putting it off and saying it’s acting outside of the means. But we have laws around these types of regulation. We’re not seeing that yet. And one last thing on when it comes to these companies, you need to think about what they want. They they want to make a lot of money. And they. And how does that not happen? Competition. And if they’re sued. So regulation prevents things like open weight models from being more of a force in the industry right now. Yes. The best models and the prices will come down, but models are heavily subsidized. We used to see a tokenization race where companies were encouraging their employees use as many as possible. Possible tokens are expensive. And so when we stop subsidizing these, these different you can get different tiers of a subscription. Openai or open weight becomes way more attractive. So I think they want to they’re wary of open weight models. And I think they’re also wary of being sued. And regulation protects them. So the next time openai or anthropic, nothing protects them from a rogue AI agent breaking into,"