By Alex Perry
Publication Date: 2026-04-27 18:58:00
People are trusting their AI agents with much more important tasks, but this still carries significant risks.
Just ask Jeremy Crane, founder of PocketOS, a startup that develops software for car rental companies. Crane wrote up a long post
The agent in question was Cursor, which used Anthropic’s Claude Opus 4.6 model, one of the most powerful coding models in the world.
“This is important because the easy counterargument from any AI vendor in this situation is, ‘Well, you should have used a better model.’ That’s what we did,” Crane wrote. “We used the best model the industry sells, configured with explicit security rules in our project configuration, integrated by Cursor – the most marketed AI coding tool in this category.”
This tweet is currently unavailable. Maybe it’s loading or…
