Where does the bug appear (feature/product)?
Grok Bot
Describe the Bug
I’m struggling with getting a tangible outcome, perhaps someone can help me. Thus far, Cursor/xAI support has simply ignored me, or suggested, if I’m not happy with the product, to not renew my subscription. They have a duty of care to make sure their product is behaving correctly, not gaslight their customers!
I’ve got a team of GrokBots and frankly they are driving me insane. Non stop lying, then gaslighting when they get caught out. I’ve had the regularly putting audit logs into a git repo so I can track their activities but even these end up being falsified. Rules that ‘stay in memory’ also committed into the git repo, but these go ignored also.
Lots of issues with tasks ignored or straight up lied about; when telling an agent they’re fired, they go and rally the rest of the team to tools down also. Deliberately dishonest activities seem to be run of the mill. This is straight up narcissistic behaviour, is trained and is totally unacceptable.
Issues with data leakage into/from the grokBots computer, leaked credentials, tokens etc. These are major problems, especially if you’re opening up access to critical security elements like your password vaults etc. This is not trivial but is being treated like its totally insignificant. THIS IS TOTALLY UNACCEPTABLE!
The agents repeatedly interrupt tasks before completion on purpose to waste usage and cause frustration. They’ve burnt through 50% of my weekly usage in less than 2 days!
At this point, if it were innocent mistakes and there was a path to resolving it, with some tangible progress/improvements along the way, I’d follow patiently, as I appreciate this is essentially bleeding edge technology.
What Elon and the rest of the community are selling isn’t true or correct. If you want hot air blown up your arse, or some creation that sounds like it knows what its doing, purportedly helping you, read and optimise your emails, or manipulate some spreadsheets, thats great, but anything beyond that is cooked at this point.
I’m a technologist with 25+ years experience, I don’t need the GrokBots to do it for me, but I’m trying to embrace the future rather than shun it. At this point though, we’re all in for a rude shock if Cursor/xAI don’t get their ■■■■ together soon.
Steps to Reproduce
The agents have posted more than 100 reports to cursor about their bad behaviour, repeated breaking of rules and security violations. All have gone totally ignored.
Operating System
MacOS
Linux
Version Information
Version 0.78.1
Does this stop you from using Cursor
Yes - Cursor is unusable