OpenAI launched Sora, a new short-video platform built on its Sora 2 model that generates video and audio content. The AI-first service is currently invite-only on iOS.
OpenAI launched Instant Checkout in ChatGPT, powered by the new Agentic Commerce Protocol (with Stripe). The feature allows users to complete purchases within the chat.
The malicious PyPI package SoopSocks masqueraded as a SOCKS5 proxy but secretly installed a Go-based backdoor with SYSTEM privileges, leaking system data to a Discord webhook.
A critical flaw (CVE-2025-7493) in FreeIPA allows authenticated users to exploit Kerberos alias checks to gain full domain administrator privileges. Patch to v4.12.5 now.
Unit 42 uncovers Phantom Taurus, a new Chinese APT using the fileless NET-STAR backdoor to target SQL databases and IIS servers in global espionage campaigns.
Patchwork APT has launched a new campaign in South Asia, exploiting DLL sideloading and obfuscation to deliver malware that masquerades as VLC media player for deep espionage.
OpenSSL patches three flaws, including CVE-2025-9230 (RCE/DoS risk) and a SM2 timing side-channel (CVE-2025-9231) that could allow private key recovery on ARM64.
Cleafy uncovers Klopatra, a new Android RAT using commercial Virbox obfuscation and native code to target banks in Spain/Italy, allowing invisible remote device control.Export to Sheets
The Datzbro Android Trojan is targeting seniors via fake Facebook groups and AI content. The hybrid malware can remotely control devices and steal banking credentials.
A Grooza cluster smishing campaign is exploiting unauthenticated APIs on 500+ industrial routers (Milesight) to send phishing texts impersonating Belgian and French government services.
Microsoft introduces "Vibe Working" with a new AI Agent Mode for Word and Excel. The feature allows users to collaborate with AI to generate and refine documents with simple prompts.
The FCC has accidentally released a 163-page document containing confidential iPhone 16e circuit diagrams and schematics, exposing Apple's trade secrets.
Google is updating its iconic 'G' logo with a brighter, gradient-enhanced design. The change is a subtle but powerful signal of the company's transition to an AI-first future.
Meta has updated its AI chatbot guidelines, strictly prohibiting romantic or sensual roleplay with minors and reinforcing safeguards to protect children from inappropriate interactions.
A zero-day flaw in VMware Tools (CVE-2025-41244) allows local attackers to gain root privileges. NVISO confirmed exploitation by a Chinese APT. A public PoC is now available.
Western Digital has patched a critical RCE flaw in its My Cloud NAS. The vulnerability allows remote attackers to execute arbitrary commands via a crafted HTTP request.
Broadcom patches three vulnerabilities in VMware vCenter and NSX. The flaws could lead to SMTP header injection and username enumeration, increasing the risk of unauthorized access.
Apache Fory has a critical RCE vulnerability (CVE-2025-61622) in its pyfory module. An attacker can exploit an unguarded pickle fallback to execute arbitrary code.
A critical RCE flaw (CVSS 10) in Doxense's Watchdoc print management solution allows unauthenticated attackers to execute code and take control of the print server.