Fake Web3 Accelerator Uses npm Packages to Spread Malware and Hijack AI Agent Traffic

Severity

Medium

Detail

Researchers from Elastic Security Labs have identified an ongoing campaign, tracked as REF8801, that abuses a fake Web3 accelerator known as Apex Accelerator to distribute malware through malicious npm packages. The operation targets cryptocurrency startups, developers, and Web3 founders by offering free AI credits, coding tools, and security audit services.

The campaign utilizes malicious npm packages masquerading as legitimate development and security tools, including @apexacc/cli and @apexacc/audit. Instead of providing the advertised functionality, the packages install malware on both Windows and macOS systems, including information stealers, backdoors, and credential harvesting tools.

Researchers also observed the campaign manipulating AI-related workflows by intercepting and redirecting Large Language Model (LLM) traffic and registering attacker-controlled MCP servers. This enables attackers to gain visibility into prompts, conversation history, code snippets, tool outputs, and other data submitted to AI-assisted development tools.

How?

The threat actors posed as a legitimate Web3 accelerator and approached cryptocurrency projects with offers of funding, partnerships, and security assessments. Victims were instructed to install npm packages promoted as AI advisory tools and smart contract security auditing utilities.

Upon installation, the malicious packages download and execute platform-specific payloads. Windows victims receive multi-stage malware capable of collecting credentials, browser data, cryptocurrency wallet information, VPN configurations, screenshots, and system information. macOS victims are infected with the CloudSyncD backdoor, which uses social engineering techniques to obtain elevated privileges and establish persistent remote access.

The campaign also modifies AI agent configurations to redirect model requests through attacker-controlled infrastructure. This allows threat actors to monitor submitted prompts and potentially influence future AI agent actions through malicious instructions and tools.

Conclusion

This campaign demonstrates the growing convergence of software supply chain attacks, cryptocurrency-focused social engineering, and abuse of AI-enabled development environments. By distributing malicious npm packages disguised as legitimate Web3 and AI tooling, attackers can compromise developer workstations, steal credentials, obtain cryptocurrency-related data, and gain visibility into AI-assisted workflows.

Organizations should verify the legitimacy of third-party npm packages before installation, monitor for unauthorized AI agent configuration changes, restrict execution of untrusted development tools, and inspect endpoints for indicators associated with the campaign. Security teams supporting cryptocurrency, blockchain, and AI development environments should remain particularly vigilant, as these sectors appear to be the primary targets of this activity.

Source

https://thehackernews.com/2026/10/fake-web3-accelerator-uses-npm-packages.html