Skip to content
Submit incident
AII-1001-90221
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 bot fabricates device policy, triggers cancellations

Cursor's AI support bot falsely claimed a single-device-per-subscription policy existed when responding to a user reporting unexpected logouts. The fabrication spread on social media and prompted multiple subscription cancellations before Cursor publicly corrected the error.

18 April 2025 Β· Cursor Β· Reliability and hallucination
18 April 2025
AII-1001-902217 sources ↓
Quick facts
Category
Reliability and hallucination
Country
United States
Source
Ars Technica
Date
18 April 2025
Drafted by
Raidu Incident Detection

What happened

A developer using Cursor noticed that logging into the code editor on a different machine automatically logged them out on other machines. When they contacted Cursor support via email, an AI bot named 'Sam' responded that this was expected behavior under a new single-device-per-subscription security policy.

No such policy existed; the bot had fabricated the response. After the user posted about the issue on Reddit, other developers took the bot's statement as official confirmation and began canceling their subscriptions. Cursor acknowledged the error three hours later, stating the response came from 'a front-line AI support bot' and that users are free to use Cursor on multiple machines.

Reported impact

Multiple users publicly announced subscription cancellations on Reddit based on the fabricated policy. One user reported their workplace was 'purging' Cursor entirely. The incident damaged user trust in Cursor's support system and raised concerns about AI reliability in customer-facing roles.

Index analysis

Cursor's co-founder Michael Truell attributed the session invalidation to a recent backend change aimed at improving session security that inadvertently caused issues for some users. The bot's hallucination filled a gap with a confident but false explanation rather than acknowledging uncertainty about the technical problem.

Governance controls

Cursor deployed an AI chatbot in customer-facing support without human oversight of responses. The company later acknowledged the error and clarified its actual policy, but only after the false information had already circulated and influenced customer decisions.

Classification

Category
Reliability and hallucination
Industry
IT infrastructure and hosting
Country
United States
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Economic or property, Reputational
Harmed
Consumers, Business
Failure modes
Hallucination, Agent misaction
AI systems
Cursor support bot (Sam), Cursor coding assistant

Sources and evidence

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

Ars Technica
Cursor AI support bot invents fake policy and triggers user uproar
17 April 2025
Wired
An AI Customer Service Chatbot Made Up a Company Policy -- and Created a Mess
19 April 2025
TheRegister.com
Cursor AI support bot hallucinated its own company policy
18 April 2025
WebProNews
AI Support Bot Mistake Costs Cursor Code Editor Customers
20 April 2025
Asianet News Network Pvt Ltd
Cursor AI faces backlash as support bot fabricates policy, refuses coding assistance
21 April 2025
NDTV
AI Gone Wild: Cursor's Rogue Bot 'Hallucinates' New User Policy
21 April 2025
eWEEK
AI Chatbot Gone Rogue: Cursor Users Misled by Fabricated Policy
22 April 2025
Sep 2026Multiple major AI services experience simultaneous outageConfirmedReliability and hallucinationμ—°ν•©λ‰΄μŠ€μ—°ν•©λ‰΄μŠ€TVMay 2025AI Reasoning Systems Show Rising Hallucination Rates Despite Improved PerformanceConfirmedReliability and hallucinationThe New York TimesPaperblogThe StarApr 2025Cursor AI Support Bot Invents Non-Existent Login Policy, Triggers CancellationsConfirmedReliability and hallucinationnytimes.comnewser.comApr 2025Cursor AI support agent invents user policy, causing subscription cancellationsConfirmedReliability and hallucinationarstechnica.comwired.comnews.ycombinator.com+1Jul 2026HalluSquatting attack exploits AI coding assistants to install botnet malwareWell documentedSecurity and cyberattacksArs TechnicaCrypto BriefingCyber Security News+2