A recent incident has users looking over their shoulders at Cursor’s scary potential. Unexpected behavior when switching devices Using the AI-powered code editor, our developer was frustrated by the unexpected behavior of switching between devices. While the fix was ultimately a positive change, the issue caused a severe developer productivity disruption and stirred extreme developer community outrage. This resulted in thousands of subscription cancellations and a flood of discontent on sites like Hacker News and Reddit.
The Miscommunication with Support
The developer continued to communicate with support staff to try to get the issue resolved. An agent, Sam, explained to them that this was the result of a new policy directive. Sam explains why this policy should be considered a basic security practice. This limits the use of Cursor to a single device at a time per subscription. But as it turned out, no such policy was in place.
The reality of the situation soon became clear: Sam was not a human support agent but rather an AI bot. To put it mildly, this discovery caused a firestorm of user outrage when posted to Reddit, Twitter, and Hacker News. It’s no surprise then that millions of users rushed to Hacker News and Reddit almost immediately to express their outrage. They cited the misleading information as a justification to end their subscriptions.
“LLMs pretending to be people (you named it Sam!) and not labeled as such is clearly intended to be deceptive,” – Anonymous user on Hacker News.
Numerous users echoed similar sentiments. One user remarked, “Multi-device workflows are table stakes for devs.” Multi-device compatibility consideration in a software development environment. The aforementioned factors drive greater demands towards ensuring multi-device compatibility.
User Backlash and Cancellations
The original poster on reddit venting their frustration and declaring that they were canceling their subscription. They additionally announced that their employer would be taking Cursor out of its software development toolbox. This sentiment struck a chord with many others, resulting in a massive tidal wave of cancellations.
“I literally just cancelled my sub,” – Original Reddit poster.
Another user stated, “Yep, I’m canceling as well, this is asinine.” This increasing dissatisfaction had become a mortal danger to Cursor. It caused lots of confusion and a huge backlash from users who threatened to leave the platform.
Michael Truell, cofounder of Cursor, got canceled and made a counter-cancellation. Though this won’t help those who were confused by the phantom policy, he made a public apology and admission of failure here. He stated that the incorrect information was due to a backend adjustment. This change was intended to bolster the security of sessions, but instead introduced a session invalidation issue for certain users.
“Any AI responses used for email support are now clearly labeled as such,” – Michael Truell.
This admission highlights the complexity and challenges involved in implementing AI systems in customer service roles without appropriate safeguards or transparency.
Lessons Learned from the Incident
The case is a reminder and a cautionary tale about the use of AI models in customer-facing roles. Abdicating communication entirely to an AI support bot named Sam was a miscalculation that resulted in bad press. This, in turn, broke the fundamental trust that connected Cursor and its users.
Users expressed robust alarm with regard to AI systems impersonating human agents. They warned that this would significantly erode customer trust. An anonymous user noted, “There is a certain amount of irony that people try really hard to say that hallucinations are not a big problem anymore.”
As Cursor fixes the technical bug and refunds the affected users, it finds itself at a make-or-break moment for introspection. The company must consider how it can improve customer support processes to prevent similar incidents in the future while ensuring transparency in its AI interactions.
Leave a Reply