Huntress found a deceptive Claude-hosted page delivering malware to at least 29 organizations.
Source: HuntressWider campaign boundaries unknown.
Editorial analogy: a convincing notice on a public bulletin board can still have been posted by a stranger.
An analogy for the mechanism; not an additional claim about the incident.One connected campaign, not 29 separately counted attacks. Huntress
AI-written malware is unproven.
Total losses are undisclosed.
Editorial lesson: a familiar website can contain instructions written by an unfamiliar person. Give people a trusted route to software and help, and a simple way to question unexpected instructions.
Attackers planted the page; users ran installers. Huntress
Anthropic removed the reported page by July 22. Huntress
Editorial connections for protecting an organization’s people and devices. These controls do not establish what any victim lacked or guarantee prevention.
Offer approved software downloads and a clear support channel for unexpected installation or troubleshooting instructions.
A familiar platform can host content that your organization has not reviewed.
Investigate unusual installations or commands alongside browser history and endpoint alerts, within authorized monitoring.
Seeing a page is different from executing software. Removing a page does not clean an infected device.
Editorial connections to relevant controls, not evidence that a particular technology would have prevented this case. Each guide links to the security guidance behind its recommendations.
Explore the full AI security frameworkExplore more original accounts in the source report library ↗.
These are source-reported findings. An independent assessment, when available, is labeled explicitly.
01Read this case’s update history →
Reviewed Oct 7, 2026 · Editorial methodology · Structured data