zlacker

>>scared+(OP)
(Cursor cofounder)

Apologies - something very clearly went wrong here. We’ve already begun investigating, and some very early results:

* Any AI responses used for email support are now clearly labeled as such. We use AI-assisted responses as the first filter for email support.

* We’ve made sure this user is completely refunded - least we can do for the trouble.

For context, this user’s complaint was the result of a race condition that appears on very slow internet connections. The race leads to a bunch of unneeded sessions being created which crowds out the real sessions. We’ve rolled out a fix.

Appreciate all the feedback. Will help improve the experience for future users.

>>mntrue+1F4
so the actual implementation of the code to log people off was also hallucination? the enforcement too? all the way to a production environment? is this safe, or just a virtual scape goat?

>>ach9l+o95
To my understanding there weren't really distinct "implementation of the code to log people off" and "enforcement" - just a bug where previous sessions were being expired when a new one was created.

That an LLM then invented a reason when asked by users why they're being logged out isn't that surprising. While not impossible, I don't think there's currently indication that they intended to change policy and are just blaming it on a hallucination as a scape goat.