Threat Feed
Daily intelligence on vulnerabilities, threat actors and geopolitical context — distilled from primary sources.
Filtered Reports
30 / 35 results
highperson_alertThreat ActorUNC6293, UNC7005, UNC5976 Abuse OAuth and WhatsApp Linking for Espionage
UNC6293, UNC7005, and UNC5976 are three distinct suspected Russian cyber espionage threat clusters conducting persistent account compromise operations. UNC6293 is assessed to be a sub-cluster of Ice Relic (formerly APT29, also tracked as Cozy Bear an…
highbug_reportVulnerability737 fake Chrome VPN extensions route traffic through attacker SOCKS5 proxies
Google Chrome Web Store users who installed any of 737 malicious extensions impersonating VPN/proxy services (Proton VPN, NordVPN, Surfshark, ExpressVPN, Cloudflare 1.1.1.1). Approximately 75,000 downloads recorded, primarily Russian users.
highbug_reportVulnerability737 malicious Chrome VPN extensions route traffic through attacker proxies
Google Chrome users who installed any of 737 malicious VPN/proxy extensions from Chrome Web Store, primarily targeting Russian-speaking users. 274 extensions impersonated 66 legitimate VPN brands (Proton VPN, NordVPN, Surfshark, AdGuard VPN, Browsec,…
highbug_reportVulnerabilityOpenAI, Anthropic, Google reasoning APIs leaked secrets via session replay
OpenAI, Anthropic, and Google reasoning APIs (GPT-5.6 Luna, Claude Haiku 4.5, Gemini Robotics ER-1.6). Affects developers who published raw agent logs containing encrypted reasoning objects.
highbug_reportVulnerabilityCSS attacks bypass webmail sanitizers to steal passwords and tokens
Microsoft Outlook, Gmail, Fastmail, Proton Mail, Yahoo Mail, and AOL Mail webmail interfaces. Attacks exploit CSS and HTML parsing discrepancies to escape message boundaries.
highbug_reportVulnerabilityAWS, Google, Vercel agent flaws allow tool execution without model checks
Amazon Bedrock AgentCore InvokeHarness API (fixed July 31, 2026; CVE-2026-18830, CVSS 8.6), open-source Strands Python library (unpatched resume path remains), Google Agent Development Kit (ADK) for Python <2.5.0 (CVE-2026-18236, CVSS 9.3), Vercel AI…
highbug_reportVulnerabilityGoogle removes ADK workflows after prompt injection exposed CI credentials
Google Agent Development Kit (ADK) Python repository on GitHub. Three workflows removed: issue-analyze.yml, issue-fix.yml, and pr-analyze.yml. Affected repository automation infrastructure, not the distributed ADK Python package itself.
highbug_reportVulnerabilityMalware can hijack Google Password Manager passkeys on Windows via TPM abuse
Google Password Manager synced passkeys on Chrome for Windows with TPM. All three attacks require pre-existing malware on the victim's Windows device. Services that do not properly validate user verification flags (e.g., eBay, now patched) are vulner…
highbug_reportVulnerabilityChrome Password Manager passkey bypass allows malware to hijack accounts
Google Chrome Password Manager on Windows systems with TPM. All three attack paths require malware already running as an ordinary user. Specific affected Chrome versions not disclosed.
highbug_reportVulnerabilityChrome 149–151 fix 1,442 flaws as AI-driven bug discovery outpaces patching
Google Chrome versions 149, 150, and 151 (released June–July 2026). All prior Chrome versions are affected by the resolved vulnerabilities. One critical flaw (CVE-2026-3545, CVSS 9.6) is a 13-year-old sandbox escape in Navigation component, patched i…
highperson_alertThreat ActorChaos Ransomware Group Deploys msaRAT Rust Implant via Headless Browsers
Chaos is a ransomware group that operates through spam floods, vishing campaigns, Quick Assist abuse, and RMM tools for initial access and persistence. The group deploys custom tooling including the msaRAT Rust-based implant as a pre-ransomware stage…
highperson_alertThreat ActorChaos ransomware gang deploys msaRAT backdoor via browser hijacking
Chaos is a ransomware gang that emerged in early 2025, distinct from the earlier same-named ransomware family active since 2021. The group has been linked to Iranian state-backed threat actor MuddyWater, who reportedly leveraged Chaos ransomware to d…
highbug_reportVulnerabilityAdobe Acrobat Chrome extension flaw exposed WhatsApp Web chats
Adobe Acrobat extension for Chrome versions 26.5.2.1 and below. Affects approximately 329 million browser installations. Exploitation requires victim to visit attacker-controlled webpage while extension is installed and WhatsApp Web is in use.
highbug_reportVulnerabilitySandbox escape flaws in Cursor, Codex, Gemini CLI, Antigravity AI tools
Multiple AI development tools: Cursor IDE, OpenAI Codex, Google Gemini CLI, and Antigravity. Vulnerability affects AI agent sandbox implementations where agents write files executed by host tools, allowing escape from restricted environments.
criticalbug_reportVulnerabilityGoogle Dialogflow CX flaw lets attackers hijack agents in same GCP project
Google Dialogflow CX agents with Code Block feature enabled, within shared Google Cloud projects. Requires attacker to have edit rights on at least one agent in the project.
highbug_reportVulnerabilityPhishing campaign targets marketing professionals via fake job interviews
Marketing professionals with Google accounts; campaign impersonates 30+ brands including Adobe, Netflix, Coca-Cola, OpenAI. Credential theft targeting Google accounts specifically.
highperson_alertThreat ActorNetNut Residential Proxy Network Disrupted After Compromising 2M Devices
NetNut operated a residential proxy network that leveraged approximately 2 million compromised Android devices to provide unauthorized proxy services. The actor monetized access to infected devices including smart TVs and streaming boxes, selling res…
highperson_alertThreat ActorNetNut (Popa) Residential Proxy Botnet Disrupted by Google, FBI
NetNut, also tracked as Popa, operates a residential proxy network built on approximately 2 million compromised home devices. The actor monetizes this infrastructure by selling proxy services that route malicious traffic through legitimate residentia…
highperson_alertThreat ActorToddyCat Deploys Umbrij Malware to Hijack Gmail via OAuth Abuse
ToddyCat (G1022) is an advanced persistent threat group that has demonstrated sophisticated capabilities in targeting corporate and enterprise environments.
highbug_reportVulnerabilityFake Perplexity AI Chrome extension hijacks search traffic on Web Store
Google Chrome users who installed the malicious Perplexity AI impersonator extension from the Chrome Web Store. Affects organizations and individuals using Chrome browser seeking AI productivity tools.
highperson_alertThreat ActorSilent Swap Campaign Targets Cryptocurrency via Malicious Browser Extensions
Silent Swap is an active cybercrime campaign leveraging malicious browser extensions to conduct cryptocurrency theft. The operation distributes unsigned installers written in both .NET and Golang variants, masquerading as a legitimate Google Notes br…
highbug_reportVulnerabilityAirDrop and Quick Share flaws enable wireless DoS and security bypass
Apple AirDrop and Google Quick Share wireless file transfer features on iOS, macOS, and Android devices. Specific affected versions not disclosed. Attack requires physical proximity (wireless range).
highperson_alertThreat ActorMalicious Chrome Extension Impersonates Perplexity AI to Intercept Searches
The threat actor behind this campaign remains unattributed. The operation demonstrates a financially or espionage-motivated adversary leveraging social engineering through brand impersonation of Perplexity AI, a popular search technology.
highbug_reportVulnerabilityAdblock for YouTube extension with 10M+ installs contains code injection risk
Chrome extension "Adblock for YouTube" (10+ million active installations). All users with the extension installed are potentially affected. Extension currently holds Featured badge status in Chrome Web Store.
criticalbug_reportVulnerabilityCI/CD flaw "Cordyceps" exposes 300+ GitHub repos to supply-chain takeover
300+ GitHub repositories across major organizations including Microsoft, Google, and Apache. Vulnerability affects GitHub Actions CI/CD workflows. Specific products and versions not disclosed in available data.
highperson_alertThreat ActorRussian-speaking actor deploys OXLOADER to distribute CastleStealer
The threat actor is a Russian-speaking, financially motivated cybercrime operator leveraging malicious advertising infrastructure for initial access. The actor demonstrates capability in developing or acquiring custom malware tooling, including the p…
highbug_reportVulnerabilityGoogle Cloud Vertex AI SDK flaw enables ML model hijacking via pickle attack
Google Cloud Vertex AI SDK for Python. Specific affected versions not disclosed. Impacts organizations using the SDK to upload and deploy machine learning models to Google Cloud's serving infrastructure.
highbug_reportVulnerabilityVertex AI Python SDK vulnerable to RCE via bucket squatting attacks
Google Vertex AI Python SDK. Affects users uploading models to Vertex AI. Vulnerability exploits bucket squatting during model upload process combined with pickle deserialization to achieve cross-tenant remote code execution.
highperson_alertThreat ActorChinese Cybercrime Network Weaponizes Google Gemini AI for SMS Phishing
A Chinese cybercrime network operating a phishing-as-a-service (PhaaS) platform called Outsider. The group weaponizes Google's Gemini AI to craft and conduct SMS-based phishing attacks (smishing) targeting American victims.
highperson_alertThreat ActorUNC1151/Ghostwriter Targets Polish Gmail Accounts in Phishing Campaign
UNC1151, also known as Ghostwriter, is a threat actor group attributed by FireEye/Mandiant that has conducted sustained information operations and cyber espionage activities.