Backdoored Claude Code Extension Distributed via Cursor Marketplace After Credential Leak

Thumbnail Image

The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.

A leaked Fireworks AI access key enabled attackers to steal coding agent records and credentials, leading to the distribution of backdoored Anthropic Claude Code extensions in Cursor's VS Code marketplace. The malicious versions exposed user data and enabled unauthorized access, affecting up to 5,000 developers before remediation. Most users were unaffected; affected customers were notified.[AI generated]

Why's our monitor labelling this an incident or hazard?

The event involves an AI system (Anthropic's Claude Code extensions) that was maliciously copied and distributed, leading to unauthorized access to software tools and data. This unauthorized access and data exposure represent harm to property and user data, fulfilling the criteria for an AI Incident. The incident is not merely a potential risk but a realized harm, as affected customers were notified and the company took remediation actions. Hence, it is not a hazard or complementary information but an AI Incident.[AI generated]
AI principles
Robustness & digital securityPrivacy & data governance

Industries
IT infrastructure and hostingDigital security

Affected stakeholders
Consumers

Harm types
Human or fundamental rights

Business function:
Research and development

AI system task:
Content generation


Articles about this incident or hazard