This week, the world of cybersecurity once again demonstrated its unpredictability: from hacking government alert systems to sophisticated methods of bypassing AI analysis. Let's look at the key events that deserve close attention.

Brazilian "alien" spam: hackers paralyzed the emergency communication system

On the night of June 20, 2026, Brazil's national emergency alert system, Defesa Civil Alerta, was hacked. Attackers, having gained access to Civil Defense employee accounts, sent false high-priority warnings to residents of several states. Smartphone sirens went off even in silent mode, and the messages contained incoherent texts, slang, and even mentions of an "alien attack." To stop the spam attack, authorities had to forcibly shut down servers at 1:30 AM. This incident is a stark example of how vulnerabilities in state digital infrastructures can be exploited for large-scale psychological impact and disorganization.

Canada: first-ever warrant for remote device cleanup

The Canadian Security Intelligence Service obtained an unprecedented court warrant for remote intervention in the operation of infected servers, home routers, and IoT devices. Botnets operating on a relay scheme used outdated equipment (specifically, D-Link DIR-850L and DIR-818LW models) to scan critical infrastructure networks. Researchers from XLab discovered the AryStinger botnet, which hacked over 4,000 routers, turning them into proxy servers. This highlights a critical issue: millions of devices with factory passwords or end-of-life support become ideal targets for hackers. About 48% of infections were concentrated in South Korea, China, and Sweden.

macOS Gaslight: AI analysis fooled by fake errors

Researchers from SentinelOne discovered a new infostealer for macOS — Gaslight. Its uniqueness lies in its targeted attack on AI-based automated analysis tools. Hidden inside the file is a loader with 38 fabricated system messages that mimic memory overflow errors and crashes. These strings act as prompt injections for LLM models, causing the AI agent to abort analysis of the "corrupted" sample. This marks a new turn in the arms race between hackers and security systems.

Europol dismantles Amadey and StealC network

A coordinated operation by Europol and law enforcement from a dozen countries led to the seizure of 326 servers and 142 domains used to distribute the SocGholish, Amadey, and StealC malware. The results are impressive: crypto assets worth over $47 million were frozen, a database with 27 million stolen credentials was seized, and approximately 15,000 WordPress sites were cleaned. The Amadey trojan served as a loader for the StealC infostealer, which stole passwords, card data, and crypto wallet seed phrases. Concurrently, 69 people were arrested in Hong Kong for laundering $25.6 million through fake accounts.

ZachXBT reveals identity of hacker arrested in Poland

Polish law enforcement, with support from the FBI, arrested four members of a hacker group suspected of SIM swapping and stealing crypto assets. The attackers cloned victims' phone numbers, bypassing two-factor authentication, and withdrew funds from exchanges. On-chain investigator ZachXBT identified one of the detainees as Wojtek Kulish (nickname Merry), by matching designer items from the operational video with his Instagram. This is an excellent example of how public activity on social media can be the key to exposure.

My comment: This week showed that cybersecurity is no longer just a matter for technical specialists. Attacks on government systems, the use of IoT junk, and bypassing AI are a signal for everyone: from ordinary users to regulators. Outdated equipment and weak account protection are not just vulnerabilities, but direct gateways for large-scale attacks.