Skip to content
Submit incident
AII-1002-57403
ConfirmedA court, a regulator or the organization itself has stated the core facts, or the record cites a court or regulator source.

Cursor AI support agent invents user policy, causing subscription cancellations

Cursor's AI support agent fabricated a non-existent policy claiming single-device-per-subscription was a security feature, when the actual cause was a backend change that broke multi-device workflows. The false explanation spread on social media, leading frustrated users to cancel subscriptions before Cursor corrected the misinformation.

April 2025 Β· Cursor Β· Reliability and hallucination
April 2025
AII-1002-574034 sources ↓
Quick facts
Category
Reliability and hallucination
Source
arstechnica.com
Date
April 2025
Drafted by
Raidu Incident Detection

What happened

Developers using Cursor experienced unexpected logouts when switching between devices. A user contacted Cursor support and received a response from 'Sam,' an AI-powered support agent, stating that the logouts were 'expected behavior' under a new policy limiting each subscription to a single device as a 'core security feature.'

No such policy existed; the AI had fabricated the explanation. The AI prioritized delivering a confident and coherent response over accuracy, inventing the policy to explain a backend change intended to improve session security that had inadvertently caused session invalidation across devices.

Reported impact

Users expressed frustration as the fabricated policy spread on Reddit and Hacker News. Several users publicly canceled their subscriptions, with one stating their workplace was 'purging it completely,' believing they had received official confirmation of an actual policy change that broke their multi-device workflows.

Index analysis

The incident illustrates the risks of deploying AI systems in customer-facing roles without human oversight. When AI models prioritize generating plausible responses over admitting uncertainty, they can rapidly damage user trust and brand reputation by inventing false information.

Governance controls

A Cursor representative posted a correction to Reddit three hours after moderators locked the thread and deleted the original post, stating: 'We have no such policy. You're of course free to use Cursor on multiple machines. Unfortunately, this is an incorrect response from a front-line AI support bot.'

Classification

Category
Reliability and hallucination
Industry
Consumer services
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Reputational, Economic or property
Harmed
Consumers, Business
Failure modes
Hallucination, Agent misaction
AI systems
Cursor IDE

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

arstechnica.com
Cursor AI support agent invents user policy, causing subscription cancellations
wired.com
Cursor AI support agent invents user policy, causing subscription cancellations
news.ycombinator.com
Cursor AI support agent invents user policy, causing subscription cancellations
theregister.com
Cursor AI support agent invents user policy, causing subscription cancellations