macOS.Gaslight Rust Backdoor with Prompt Injection
Score: 8/10

macOS.Gaslight Rust Backdoor with Prompt Injection

DPRK-aligned actors are deploying macOS.Gaslight, a Rust-based backdoor that utilizes prompt injection payloads to subvert LLM-assisted malware analysis pipelines.

Executive Summary

SentinelLABS has identified a new Rust-based macOS implant, tracked as macOS.Gaslight, attributed with high confidence to DPRK-aligned activity. This malware is significant for its use of a 3.5 KB prompt-injection payload designed specifically to mislead and abort LLM-assisted triage pipelines by mimicking system failure messages and security warnings.

The implant utilizes the Telegram Bot API for command-and-control (C2), featuring hardened transport via AES-GCM and certificate pinning to frustrate network inspection. It further complicates forensic analysis by self-redacting its C2 bot token from runtime memory and logs. The malware targets browser data, keychains, and system profiles, often deploying a secondary Python-based collection module.

This campaign highlights an evolution in adversary tradecraft where threat actors are now weaponizing the tools used by defenders. As SOC teams increasingly rely on AI and LLMs for automated analysis, the inclusion of analyst-targeting prompt injection poses a direct threat to the integrity of automated security operations.

Key Details

Threat Name

macOS.Gaslight

Affects

—

Adversary

DPRK-aligned macOS activity

Malware/Tools

macOS.Gaslight, Atomic macOS Stealer, PhantomCaptcha, Hades, Shai-Hulud

Report Score

8out of 10
Quality Score
Good
IOC Quality8
TTP Details9
Detection Guidance6
Enterprise Relevance8
Clarity & Structure10
Technical Depth9

Sources