Assisted – Bitcoin Platform https://bitcoinplatform.com Breaking Crypto News and Blockchain updates Tue, 29 Sep 2026 00:18:01 +0000 en-GB hourly 1 https://wordpress.org/?v=7.1.5 https://bitcoinplatform.com/wp-content/uploads/2026/09/cropped-fevi-18-32x32.png Assisted – Bitcoin Platform https://bitcoinplatform.com 32 32 Malicious npm Dependency Linked to AI Assisted Commit Targets Crypto W https://bitcoinplatform.com/malicious-npm-dependency-linked-to-ai-assisted-commit-targets-crypto-w/ https://bitcoinplatform.com/malicious-npm-dependency-linked-to-ai-assisted-commit-targets-crypto-w/#respond Tue, 29 Sep 2026 00:18:01 +0000 https://bitcoinplatform.com/malicious-npm-dependency-linked-to-ai-assisted-commit-targets-crypto-w/ A recent discovery by researchers at ReversingLabs has unveiled a malicious npm dependency connected to an AI-assisted code commit that has been actively stealing sensitive data and exposing crypto wallets. The deceptive package, masquerading as a validation tool, allowed cybercriminals to extract secrets from compromised environments and gain access to financial assets.

The nefarious activity, known as PromptMink, revolved around the utilization of the package @validate-sdk/v2, which was integrated into an autonomous trading agent back in February of 2026. The code commit associated with this package was allegedly co-authored by Anthropic’s Claude Opus model.

The perpetuators behind this attack have been linked to the North Korean state-sponsored group Famous Chollima, also recognized as APT37 or Reaper, a threat actor notorious for targeting cryptocurrency developers. Their modus operandi involved a sophisticated two-layered approach, where seemingly legitimate tools were presented as Web3 utilities to attract unsuspecting users, while hidden malicious payloads were discreetly delivered through secondary dependencies. This duplicitous tactic enabled the attackers to maintain the facade of trustworthiness in widely visible components while surreptitiously executing their malicious agenda.

Over a span of seven months, researchers identified more than 60 packages and over 300 versions connected to the PromptMink campaign, indicating a sustained and evolving operation focused on refining their delivery techniques.

As the PromptMink campaign progressed, the malicious payload underwent significant evolution, expanding beyond mere credential theft to encompass a range of capabilities aimed at maximizing impact and persistence. These enhancements included scanning directories for crucial files and crypto-related data, harvesting system information like usernames and IP addresses, compressing entire project folders before extraction, and installing SSH keys to facilitate persistent remote access.

Furthermore, the malware underwent technical advancements, transitioning from JavaScript-based code to compiled binaries and Rust-based payloads. This strategic shift not only enhanced evasion tactics but also enabled the malware to operate seamlessly across both Linux and Windows environments.

Evidence discovered within the code indicated the utilization of large language models (LLMs) in the development process, suggesting a concerning trend where cybercriminals are tailoring malicious packages to appeal to AI coding assistants, thereby extending the reach of supply chain risks into automated development workflows.

This latest revelation serves as a stark reminder of the evolving and sophisticated nature of cyber threats, underscoring the critical importance of vigilance and robust security measures in safeguarding against malicious actors within software supply chains.

]]>
https://bitcoinplatform.com/malicious-npm-dependency-linked-to-ai-assisted-commit-targets-crypto-w/feed/ 0