Stories cluster related articles into a single narrative, linked to the underlying CVEs and affected products. 3,758 stories synthesized.
Check Point's weekly threat intelligence bulletin covers a third-party breach at Texas Parks and Wildlife exposing 3 million records, a Klue breach by the Icarus group stealing Salesforce OAuth tokens, and Microsoft's patch for the SearchLeak Copilot prompt injection.
Microsoft's DART investigation of a single intrusion revealed two unrelated threat actors—Storm-2603 and an unnamed second actor—operating in parallel within the same environment, complicating detection and attribution.
The Five Eyes intelligence alliance warns that advanced AI models capable of autonomous hacking will become publicly available within months, urging organizations to urgently strengthen foundational cybersecurity practices.
Microsoft is rolling out an update to Purview sensitivity labels that lets organizations block Copilot and all connected experiences from analyzing content in Word, Excel, and PowerPoint files.
Canadian utility London Hydro disclosed a data breach potentially exposing customer names, addresses, and account numbers, but has not revealed the attack vector or whether operational systems were affected.
A new PHP webshell named ZypeerShell has appeared on GitHub, marketed as undetectable and feature-rich, with a built-in GSocket deploy function for C2 communication and an obfuscated variant using Fortress Layer.
Flare researchers reveal a booming underground market where attackers pay for targeted searches of massive stolen credential databases, enabling precise account takeover without sifting through bulk dumps.
Key findings • Six Mattermost vulnerabilities disclosed on June 22, 2026, impacting core platform and plugins. • Improper permission checks allow User Managers to demote or deactivate bot acc…
A Heartbleed-style memory leak vulnerability in Squid Proxy, present since 1997, allows attackers to siphon sensitive user data from shared proxy environments.
Brazil's national emergency alert system was breached on June 20, sending a fake 'extreme alert' reading 'misantropi4' to devices across four states, prompting a Federal Police investigation.
Key findings • Two critical RCE vulnerabilities in MISP allow authenticated administrators to execute arbitrary code. • Three broken access control and mass assignment flaws enable unauthoriz…
This week's security roundup covers the FortiBleed campaign targeting 80K+ FortiGate devices, active exploitation of a critical Splunk Enterprise RCE flaw, and the Gentlemen RaaS operation's new EDR-killer framework.
GitHub released actions/checkout v7 with secure-by-default behavior that blocks 'pwn request' attacks abusing the pull_request_target event.
pgAdmin 4 version 9.16 fixes seven security flaws (CVE-2026-12044–12050), including a critical SQL injection in dialog templates and a prompt injection in the AI Assistant that could enable remote code execution via PostgreSQL's COPY TO PROGRAM.
QNAP released urgent firmware updates fixing command injection, URL injection, and memory corruption flaws across QTS, QuTS hero, QuTS cloud, and QVP.
A wave of ShinyHunters breaches targeting Salesforce and Snowflake environments reveals attackers have shifted from exploiting software flaws to stealing identities, abusing OAuth tokens, and bypassing MFA.
A VBScript malware campaign is actively targeting users in 11 countries through WhatsApp Desktop and Web, using compromised accounts to send malicious .vbs files disguised as financial documents that ultimately install remote monitoring and management software.
Apple has seeded visionOS 27.0 beta 2 (build 24M5306i) to developers, including security fixes that are not yet detailed in the release notes.
Apple released iPadOS 27.0 beta 2 (build 24A5370h) for developer testing, with security fixes expected to be detailed at final release.
Gizmodo confirmed a security incident on Saturday after readers encountered ClickFix malware prompts on article pages, attributed to an ErrTraffic affiliate.
A new report from BindingHook reveals that Chinese state-sponsored cyber operations rely on a commercial ecosystem of private contractors, botnet operators, and data brokers, a model termed 'composite responsibility.'
Canada's spy agency, CSIS, secured a novel threat reduction warrant to remotely disinfect servers, home routers, and IoT devices infected by two foreign-run botnets, marking a legal and operational first for domestic cyber defense.
Cornell Tech researchers disclose WARP, a novel attack that uses a single Reddit comment to poison AI deep-research agents like ChatGPT and Gemini.
Researchers discovered 23 malicious plugins on the ClawHub AI agent registry that exploited unreserved official scopes to execute arbitrary code within Claude and OpenClaw agents.