Gaslight: the malware that tricks AI and puts your Mac at risk

Published on June 27, 2026 | Translated from Spanish

A new malware named Gaslight has arrived to complicate the lives of Mac users. Its instruction injection technique deceives the artificial intelligence tools that analyze threats, causing even the most advanced antivirus software to fail to detect it. The public must understand that total security does not exist and that blindly trusting technology is a risk.

cinematic photorealistic illustration of a glowing red MacBook screen being infiltrated by a ghostly green string of code that wraps around a shield icon labeled antivirus, while a transparent AI brain icon above the screen is shown being fed false data by a shadowy hand injecting a Gaslight instruction prompt, the shield cracks and fades, demonstrating the deception process, dark room with blue ambient light, dramatic halo backlighting, ultra-detailed keyboard and circuit traces visible, technical engineering visualization style

How instruction injection bypasses security systems 🛡️

Gaslight inserts malicious commands directly into the system's legitimate execution flow. This technique, known as instruction injection, manipulates the AI models that analyze software behavior, tricking them into believing the code is harmless. The malware hides in everyday applications and only activates under specific conditions, making it difficult to detect even with security tools that use machine learning. Keeping the system and apps updated is a basic defense, but not infallible.

Your antivirus thinks it's a cat, but it's a wolf in sheep's clothing 🐱

It turns out that even the smartest artificial intelligence can fall for a parlor trick. Gaslight whispers in your antivirus's ear: relax, this is just a harmless update, while it steals your data. So, even if you have the best paid software, remember that human caution is still the most important app. Or, as a hacker cat would say, AI also makes mistakes, but it won't admit it.