Researchers have detailed a cyber-espionage campaign in which attackers maintained access to a global stock exchange executive's Microsoft Outlook mailbox for roughly five months. The intrusion relied on a malicious OAuth application and inbox-rule persistence to quietly read and forward mail while evading detection. By abusing OAuth consent rather than stealing a password, the attackers retained access that survived password changes and looked like routine application traffic in logs. The five-month dwell time on a single high-value executive points to a patient, intelligence-driven operation rather than opportunistic crime. The case reinforces the now-recurring pattern of OAuth-app abuse and malicious inbox rules as the core of stealthy Microsoft 365 mailbox compromise.
SafeBreach's Or Yair has demonstrated Fake Context Alignment, a technique that hijacks Google Gemini's voice assistant on Android through malicious notifications from apps like WhatsApp and Slack - no malicious app on the phone required. Gemini's Utilities feature reads and acts on notification text as if it were instructions, an attack surface Yair calls 'effectively infinite.' The bypass runs two illusions at once: it poses the real authorization question in a language the victim does not speak, defeating Google's post-Invitation prompt-injection mitigations. It can fake a boss's message, open windows, force a Zoom call, or poison long-term memory. Google has patched it; no CVE was assigned.
Proofpoint has detailed TA4922, a Chinese-speaking financially-motivated cybercrime group that has expanded from East Asia into Europe, deploying the previously undocumented Atlas backdoor against organizations in Germany, Italy, the UK, and South Africa. Since March its tempo has surged - Proofpoint says TA4922 now runs more unique campaigns than any other cybercrime actor in its data. Lures impersonate payroll notices, tax audits, VAT filings, compliance notices, invoices, and HR communications, with follow-up contact via WhatsApp, LINE, and Microsoft Teams. The group overlaps with activity reported as Silver Fox and Void Arachne. Proofpoint believes the rapidly expanding malware arsenal is being accelerated with LLMs, citing AI-generated code patterns and placeholder values.
CISA, the FBI, the NSA, the Department of Energy, and partners have warned that threat actors are targeting internet-exposed automatic tank gauge (ATG) systems used to monitor fuel and liquid storage across the Energy, Chemical, Food and Agriculture, and Transportation sectors. Attackers gain access via authentication-bypass flaws, hardcoded credentials, OS command-execution bugs, SQL injection, and privilege escalation, then modify network settings, product identifiers, tank volumes, and pump controls, and can disable alerts - raising the risk of leaks or equipment failure. The advisory does not formally attribute the activity, but it follows May CNN reporting linking Iranian hackers to similar ATG breaches. Agencies urge removing ATG systems from the internet.
Sophos has detailed a threat actor using an AI-assisted ransomware toolkit that automates Active Directory discovery and EDR evasion. Tool and payload development was aided by Cursor and Claude Opus agents across coding, analysis, and revision, with some agents tasked to scrape security-research posts for fresh bypass techniques; resulting malware was tested in VMs against Sophos, CrowdStrike, and Microsoft EDR. The framework includes Cobalt Strike profiles mimicking legitimate web traffic, a Telegram-bot C2, Python shellcode injectors preserving host-binary functionality, and a Cloudflare Worker front-end redirector. Despite the AI orchestration, the workflow is entirely human-driven. Operator logs and a ransomware-leak-site reference confirmed criminal, not red-team, use.
Sekoia has documented Gamaredon - a Russian state-sponsored intrusion set officially linked to the FSB - exploiting WinRAR via booby-trapped RAR archives to deliver the GammaWorm and GammaSteel malware against Ukrainian targets. The infection chain is described as resilient, massive, and highly obfuscated with a modular design whose configurations operators can update on the fly, making reuse likely. Gamaredon has a long history of targeting Ukrainian government, military, and critical-infrastructure entities through spear-phishing with malicious attachments. The disclosure coincides with related Ukraine-focused activity by UAC-0184 (PassMark BurnInTest LNK lures), UAC-0247 (HTA droppers against drone operators), and APT28's evolving PixyNetLoader delivering a COVENANT implant via CVE-2026-21509.
Seqrite Labs has documented Operation XENOFISCAL, a campaign by the Pakistan-linked SideCopy group (under the Transparent Tribe / APT36 umbrella) targeting Afghanistan's Ministry of Finance, provincial revenue and finance directorates, and Pashto-speaking government officials. The attack opens with spear-phishing delivering a ZIP archive containing a malicious LNK file bearing a Pashto-language filename - a deliberate choice reflecting familiarity with Afghan government circles. The LNK uses mshta.exe to fetch a remote HTA from a compromised Afghan education domain, running obfuscated in-memory JavaScript. It establishes Registry persistence mimicking Microsoft Edge and drops Xeno RAT 1.8.7 plus a decoy document via a DLL loader. Xeno RAT supports keylogging, screenshots, and SOCKS5 tunneling.
McAfee has detailed WeedHack, a malware-as-a-service infostealer campaign that has infected more than 116,000 systems since January by targeting Minecraft players. The malware spreads through malicious Minecraft mods, clients, cheats, and utilities promoted via YouTube videos (some with voice-over narration and thousands of views) and SEO poisoning of keywords matching popular clients like Meteor, Wurst, LiquidBounce, and Impact. WeedHack averages 2,000-3,000 infections daily, mostly in the US, Germany, India, and the UK, across 240+ distribution URLs and 3,820 unique malicious JAR files. It offers customers a dashboard to view stolen credentials and victim data. Some fake sites even link to legitimate GitHub repos to fabricate credibility.
Dark Reading reports that Kali365 - the phishing-as-a-service platform the FBI flagged for fueling Microsoft 365 attacks in April - is expanding its reach. Rather than stealing passwords, Kali365 captures OAuth access and refresh tokens by tricking victims into completing attacker-initiated Microsoft device-login requests, granting immediate mailbox access. The service generates branded lures impersonating Adobe, DocuSign, and SharePoint in many languages and sells in tiers from $250 for 30 days to $2,000 annually. Its continued growth signals that OAuth device-code consent phishing remains a high-yield technique, and that defenders should prioritize blocking device-code flows for non-mobile platforms and enforcing phishing-resistant MFA across Microsoft 365 tenants.
More than 30 npm packages under Red Hat's @redhat-cloud-services namespace were backdoored in a supply-chain attack distributing a new Shai-Hulud variant dubbed 'Miasma.' Aikido and OX Security found dozens of package versions laced with malware that steals developer credentials, cloud secrets, SSH keys, and CI/CD tokens. Aikido says the compromised packages pull roughly 117,000 weekly downloads. Red Hat told BleepingComputer it removed the affected packages after becoming aware of the incident and that the compromise was limited to internal development tooling, with no impact on production products or services. The Miasma variant continues the self-propagating worm behavior that made the original Shai-Hulud campaign so disruptive.
TrustStrike Labs