Threat Actors Abuse Claude Code Leak In GitHub Malware Campaign

In late March 2026, a simple packaging error led to the accidental exposure of Anthropic’s internal Claude Code source material.

While the company quickly addressed the exposed data, cybercriminals immediately seized the opportunity.

Within 24 hours of the incident, threat actors launched a massive social engineering campaign on GitHub, creating fake repositories to distribute credential-stealing malware disguised as the leaked AI software.

This event highlights how organizational mistakes often catalyze severe cybersecurity threats.

On March 31, 2026, a routine update to Anthropic’s npm package inadvertently included a massive 59.8 MB JavaScript source map file.

This file accidentally exposed approximately 513,000 lines of original TypeScript code, revealing the entire agentic harness that powers the terminal-based Claude Code tool.

The leak exposed unreleased features, internal model codenames, and proprietary security mechanisms designed to prevent competitors from training on API traffic.

Infection Chain and Defense Evasion

When an unsuspecting victim downloads and extracts the fake archive from GitHub Releases, they trigger a highly sophisticated, Rust-compiled dropper program.

To avoid detection by security analysts, the malware first conducts an extensive series of anti-analysis checks.

It scans the system for signs of virtual machines, sandboxes, and debugging tools. If it detects a research environment, the malicious program silently terminates itself to avoid exposure.

Interestingly, the malware also implements a unique hardware scoring system. It evaluates the victim’s graphics card, specifically targeting modern gaming PCs.

This indicates the attackers are likely prioritizing high-performance machines for cryptocurrency mining or harvesting valuable gaming credentials.

Once the dropper confirms it is running on a valid target machine, it executes an encrypted script that systematically turns off critical Windows Defender security features.

With the system’s defenses lowered, the dropper deploys its two primary payloads: Vidar and GhostSocks.

Vidar is a notorious information stealer designed to aggressively harvest browser passwords, session tokens, cryptocurrency wallets, and system data.

According to Trend Micro research, the Claude Code incident serves as a stark reminder that human error remains a primary driver of security breaches.

Date	Incident	Description	Key Details
February 2026	AI tool lures	Malware campaign using fake AI tools	TradeAI.exe, 18+ unique samples (Copilot, Cursor, AI tools), Active campaign
March 31, 2026	Source code leak	Accidental exposure of source code	Anthropic npm packaging error, 59.8 MB source map exposed, 512K lines TypeScript
March 31 to April 1, 2026	Time window	Delay between leak and weaponization	Within 24 hours of the leak

Organizations must strictly enforce software installation policies, ensuring that developers only download AI tools from officially verified channels.

Furthermore, deploying advanced endpoint detection systems capable of identifying Rust-compiled droppers and behavioral anomalies is essential to defend against these fast-moving social engineering campaigns.

Follow us on Google News , LinkedIn and X to Get More Instant Updates. Set Cyberpress as a Preferred Source in Google.

The post Threat Actors Abuse Claude Code Leak In GitHub Malware Campaign appeared first on Cyber Security News.