On August 26, 2025, Citrix and CISA warned of active exploitation of CVE-2025-7775 in NetScaler ADC/Gateway appliances causing remote code execution and DoS. The CVSS 9.2 vulnerability requires immediate patching to prevent compromise.

On August 26-27, 2025, Google patched Chrome to fix CVE-2025-9478 use-after-free and CVE-2025-5419 zero-day in V8 engine enabling remote code execution. Users are urged to update immediately to prevent exploitation.

Zscaler ThreatLabs reported 77 malicious Android apps downloaded over 19 million times, spreading Joker and Anatsa banking trojans targeting Germany and South Korea. Users are advised to uninstall suspicious apps and use Play Protect.

On August 27, 2025, a cache deception attack was documented exploiting URL processing differences to make CDNs cache protected resources, enabling attackers to retrieve sensitive data like session cookies.

On August 26, 2025, researchers identified Hook v3, an Android malware combining ransomware, spyware, and banking trojan functions with 107 remote commands and phishing overlays. It spreads via fake sites and GitHub, stealing credentials and financial data.

Musashino Red Cross Hospital backs up 100TB+ data and achieved 500GB recovery in 20 seconds during drills using Cohesity-HPE integrated solution, improving cyberattack countermeasures and BCP.

In July-August 2025, multiple LLM security incidents including ChatGPT data leaks and code editor RCEs were reported; AI pentesting approaches emphasize continuous risk assessment and mitigation.

Proofpoint's 2025 Voice of the CISO report finds most CISOs expect material cyberattacks within 12 months, with two-thirds experiencing sensitive data loss often linked to departing employees.

Yoav Regev highlighted 'data blindness' as a major risk to AI and security due to lack of data visibility in cloud environments, urging continuous data scanning and classification.

An Italian e-commerce site with 500+ monthly orders has its admin access sold by SinCity on the dark web, risking theft of customer data including PayPal and credit card payments.

Security researchers disclosed four critical vulnerabilities in Securden Unified PAM, including CVE-2025-53118 authentication bypass enabling unauthorized access and password theft. Patches are available in version 11.4.4.

On August 26, 2025, CISA released advisories for critical vulnerabilities in INVT VT-Designer, Schneider Electric Modicon M340, and Danfoss AK-SM drives with CVSS scores up to 9.1, urging immediate mitigations.

A critical stored XSS vulnerability (CVE-2025-50975) in IPFire 2.29's firewall.cgi allows authenticated admins to inject persistent JavaScript, risking session hijacking and unauthorized internal network actions.

A critical Server-Side Request Forgery vulnerability (CVE-2025-54370) in PhpSpreadsheet allows malicious HTML injection during spreadsheet processing, posing high risk across multiple versions.

In late August 2025, CVE-2025-9411 (SQLi), CVE-2025-9416 (XSS), and CVE-2025-9429 (XSS) were reported in ruoyi-go, samarium, and mblog respectively, all with medium severity scores.

Seceon launched a cost-effective cybersecurity solution using AI/ML and Dynamic Threat Models, achieving up to 80% cost reduction and 90% fewer false positives.

ANY.RUN introduced Threat Intelligence Lookup, a searchable database aggregating malware investigation data from 15,000 SOC teams to improve alert triage and detection using MITRE ATT&CK framework.

Google revealed a PRC-linked UNC6384 campaign using social engineering and signed SOGU.SEC backdoor malware to target Southeast Asian diplomats since 2023. The attacks employ adversary-in-the-middle techniques and fake plugin updates for espionage.

Researchers uncovered ZipLine, a social engineering campaign using company contact forms to deliver MixShell malware to US supply chain manufacturers. The shellcode backdoor uses DNS TXT tunneling for covert command and control.

On August 26, 2025, attackers used phishing emails impersonating Zoom and MS Teams to deliver ScreenConnect remote access malware, targeting over 900 organizations mainly in US, Canada, UK, and Australia.

On August 26, 2025, researchers revealed ONEFLIP, an inference-time backdoor attack flipping a single bit in DNN weights to trigger malicious behavior with 99.6% success, threatening AI systems like autonomous vehicles.

ESET discovered PromptLock ransomware using AI to create Lua scripts for file encryption across Windows, Linux, and macOS, employing SPECK 128-bit encryption and written in Golang.

Mustang Panda, a China-based APT, had its cyber espionage tactics leaked; early 2025 saw dismantling of PlugX malware on over 4,200 devices distributed via malicious USB drives.