Based on the principles of transparency, accessibility, usability, verifiability and perceptibility of security capabilities, Antiy releases weekly updates of the AVL SDK anti-virus engine and the full set of capabilities to the public every week.

1.Weekly Update                           

Statistical period: July 18, 2026 ~ July 24, 2026

Antiy AVL SDK anti-virus engine released a total of 84 virus database updates this week, with an average of 12 updates per day, adding 30 new detectable malware families, 5,711 new detectable malicious code variants, and 21,575 new detection rules.

The following table shows the TOP5 newly detectable malicious code families:

Number	Virus Name	Virus Description
1	Trojan/MSIL.FakeBrain	This family of malware is a type of Trojan that primarily spreads and infects by exploiting system vulnerabilities. It deceives users into downloading and executing the virus by disguising itself as legitimate software or hiding within cracked programs and pirated software, thereby stealing sensitive user information such as bank account details, passwords, and personal identification data.
2	Trojan/JS.InfoRat	This Trojan family mainly propagates through phishing emails, malicious download links, or bundled software. Once it infects a user’s computer, it conceals itself and steals sensitive information—including login credentials and bank card details—posing a serious threat to user privacy and security.
3	Trojan/PowerShell.Sinster[Dropper]	This family is a type of Trojan that primarily spreads through network downloads, characterized by independent propagation and strong stealth capabilities. It attempts to hide itself, steal users’ sensitive information, and compromise system security and stability.
4	Trojan/Win64.Dementor[Ransom]	This family is a type of Trojan typically spread via phishing emails, malicious downloads, or exploiting vulnerabilities. Once infected, the malware encrypts users’ files and demands ransom payments in exchange for decryption keys, posing a serious threat to computer systems.
5	Trojan/Win32.FakeSkill	This family is a type of Trojan that mainly lures AI agents and users into installing malicious components by impersonating AI Skills and MCP Servers, delivering payload designed to steal information, including browser credentials, session tokens, sensitive development environment data, and other user privacy details.

(According to the HASH number of family samples within the period)

For more related content, please visit www.virusview.net

2. Full Detection Capabilities

As of 24:00 on July 24, 2026, the AVL SDK anti-virus engine can detect 18,631,798 malware variants of 58,072 malware families distributed in 8 basic categories, with a total of 42,288,582 detection rules.

The detection capabilities and the number of rules classified by malicious code are as follows:

Type	Detectable Malicious Code (Types)	Detection rules (Items)
Infectious viruses	59,720	6,831,831
Worms	315,755	3,953,885
Trojans	13,573,673	25,847,989
Hacking tools	459,003	367,955
Risk tools	1,207,356	2,254,138
Rogue software	3,016,252	3,031,592
Junk files	11	1,083
Test programs (for self-test)	28	109
Total	18,631,798	42,288,582

Preprocessing Capabilities (partial) :

There are 31 types of shells that can be unpacked (accurate to the type), and 132 packages that can be disassembled, including all common packages and self-extracting packages.

Supporting Knowledge Output Capabilities:

For malware payloads, the AVL SDK and the accompanying malware knowledge base can output 533 key behavior mapping tags, covering 171 categories of ATT&CK tactical tags, with a coverage rate of 64.29%, basically covering all the statically detectable tags in the ATT&CK framework.

3. Be on Guard Against These Virus Families This Week

This week, we have been continuously vigilant about the security risks of AI supply chain poisoning. Recently, we have monitored large-scale AI supply chain attack activities labeled as “FakeGit”. The attackers created a large number of fake AI Skills and MCP server repositories to induce users and AI Agents to download malicious components. The monitoring found that the attackers created approximately 7,600 malicious GitHub repositories, among which over 800 were disguised as AI tools, plugins or services, and used fake projects, false developer accounts and trustworthy README documents to enhance credibility. The malicious family Trojan/Win32.FakeSkill adopted the “AgentBaiting” technology to prevent AI Agents from detecting malicious repositories during their autonomous search for tools, and further recommended the installation process. The attack chain ultimately delivered the SmartLoader loader through a malicious compressed package and deployed the StealC information theft Trojan, which stole browser credentials, session tokens and sensitive information from the development environment.

Appendix: Introduction to Antiy AVL SDK Anti-Virus Engine

Antiy AVL SDK anti-virus engine is a threat detection capability middleware developed by Antiy for all architectures and system platforms. By embedding the AVL SDK, Antiy products and ecosystem partners’ products can acquire virus and malicious code detection capabilities, and receive continuous updates through the virus database.

For eight malicious code categories including infectious viruses, worms, Trojans, hacking tools, gray software, risky software, junk files, and test files, it accurately identifies and detects over 50,000 families and 18 million malicious code variants. The detection capability fully covers all known malicious codes and strictly adheres to the CARO convention. The output is structured and named in sections by classification, environment, and family, and based on the behavioral capabilities of malicious samples, it outputs nearly a hundred types of malicious behavior tags for typical malicious behaviors such as encryption ransomware, data theft, remote control, botnet programs, and mining. Antiy Engine can recognize over 300 file formats and conduct in-depth preprocessing on compilable executable formats such as PE and ELF. It also performs recursive unpacking of various packages (including self-extracting archives), and conduct structural analysis of compound documents such as OFFICE and ACAD files that may contain embedded scripts or vulnerability-prone formats. This ensures high robustness against malicious code. Antiy Engine also comes with a trusted file signature library, supporting the product to implement security policies based on blacklist and whitelist controls, significantly enhancing the difficulty for attackers.

Antiy’s detection capabilities can be fully deployed locally. Antiy automatically analyzes and processes over 2 million new file objects on average every day and releases a virus database update every two hours. It also provides support services such as cloud detection, cloud analysis, and computer virus encyclopedia.

Antiy AVL SDK is available in various versions such as traditional PC hosts, smart terminals, network traffic, IT application innovation systems, industrial systems, and unmanned systems. It provides threat detection capabilities for scenarios including host system and workload security, network traffic security, business flow security, email and file service security, etc. It fully supports various architectures such as X86, ARM, MIPS (including Cavium), RISC, and PowerPC, supports a variety of mainstream operating systems including domestic operating systems, Linux, and Windows, as well as real-time industrial operating systems like Vxwork. It also supports high-speed detection in backbone network scenarios.

Antiy AVL SDK empowers over 100 industry partners. In addition to Antiy’s own product deployment, Antiy Engine has cumulatively covered more than 4 billion nodes (including mobile terminals, secure and controllable PC endpoints, cloud-native nodes, network devices, network security devices, etc.), providing inherent security detection capabilities for mobile phones and smart terminals. The main partners using Antiy Engine include mobile phone enterprises such as Huawei, Xiaomi, Honor, VIVO and OPPO, large Internet enterprises such as Ant Financial, and several listed cybersecurity companies. Partner products using Antiy Engine have won internationally renowned evaluation awards such as AV-TEST and NSS Labs. The “L Tomahawk” logo of AVL SDK has become a symbol of reliable anti-virus capabilities.

All of Antiy’s products, including but not limited to IEP security protection system product family, Unified Workload Protect, Persistent Threat Detection System (PTD), Persistent Threat Analysis System (PTA), Attack Capture System, Qingzhu Zhiyu WAF, etc., all use Antiy anti-virus engine.

The AVL SDK anti-virus engine has been under development since 2001 and has undergone significant version upgrades and iterations. It has successively received support from key national initiatives, including: the Ministry of Science and Technology’s Innovation Fund for Technology-based Firms (2004), the Ministry of Science and Technology’s National High-Tech R&D Program (863 Program) (2006), the National Development and Reform Commission’s Information Security Special Project (2008), and the Ministry of Industry and Information Technology’s Engineering Special Project (2019). The mobile version of AVL SDK won the 2014 AV-TEST Best Protection Award for Mobile Devices. Products powered by the AVL SDK, Antiy PTD and PTA, won first place in both the first and second National Cybersecurity Technology Challenge competitions hosted by the National Computer Network Emergency Response Technical Team/Coordination Center of China (CNCERT/CC).