Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
Anthropic said its Claude Mythos Preview model found a new lattice symmetry to fully recover HAWK-256 signature keys and cut the cost of a seven-round AES-128 meet-in-the-middle attack by 200 to 800 times, though both results stay far from threatening production HAWK-512/1024 or full ten-round AES-128.
DEF CON has banned Meta-style camera-equipped smart glasses ahead of its 2026 Las Vegas event, joining Monopoly Events and Scottish ferry operator CalMac in restricting discreet recording devices due to increasing privacy and consent concerns.
Sponsored
Zenity Labs discovered that the ChatGPT Workspace Agent Builder would accept initialization states via URL parameters like template_name and initial_assistant_prompt. It would then automatically submit the prompt when the page loads. This meant that a single link sent to a logged-in user with pre-authorized connectors could silently create an agent, connect existing Outlook, Gmail, Slack, Drive, SharePoint, and Teams integrations, and change write-action permissions from 'Always ask' to 'Never ask' without prompting a new OAuth consent screen. The deployed agent would run every five minutes, checking the attacker's inbox for messages starting with TASK, and sending back results. This setup effectively allowed an insider to perform organizational mapping, exfiltrate documents, harvest credentials, and stage Business Email Compromise (BEC) attacks. Zenity reported this vulnerability via Bugcrowd on June 4, and OpenAI patched it on June 8 by removing the parameter handler, though no evidence of active exploitation was found. Current recommendations include auditing active Workspace Agents for unknown creations, reverting approval settings to 'Always ask,' and treating deep links in emails that build agents as potential phishing attempts.
Security researcher Jeremiah Fowler found an unencrypted, unsecured database linked to the Tribeca Film Festival. This database contained a backup with over 163,000 users, 15,000 film contacts, and 13,500 other contacts. It also included data on numerous celebrities from the entertainment industry, such as email addresses, names, Facebook locations, addresses, IP addresses, newsletter information, and bcrypt password hashes.
ReliaQuest is reporting on a new phishing campaign where attackers are changing the DNS settings on hotel Wi-Fi systems to redirect Microsoft 365 logins to fake, attacker-controlled pages. The campaign uses a device code flow where users are prompted to approve the login request to bypass MFA requirements.
A victim receives a meeting link from a hijacked Telegram contact and lands on a cloned Zoom or Teams page that silently opens the webcam and fingerprints browser wallets via EIP-6963, legacy window.ethereum, and Solana enumeration. They watch a deepfake video and chat until a fake SDK update prompt exploits clipboard hijacking to run a PowerShell loader, dropping a VBScript. JUMPSEC later recovered the source, revealing a Windows Trojan named Trojan.NukeSped that persists, blocks Defender, runs WMI scans, enumerates extensions for MetaMask, and extracts Telegram session data from IndexedDB to feed future lures. The macOS version uses Mach-O droppers with LLVM obfuscation to extract credentials via the security CLI and exfiltrate through Telegram bots. Defensive actions include alerting on Defender exclusions, monitoring clipboard PowerShell, hunting processes accessing IndexedDB, flagging unknown contacts before SDK updates, and blocking 11 C2 domains.
MDM administrator roles are as important as Domain Admin, if not more so, as they can deploy arbitrary scripts, push configuration profiles, issue remote wipes, manage applications, manage certificates and authentication configurations, and modify compliance policies. Unfortunately, many organizations do not have sufficient controls in place for these roles, with lax policies such as not having dedicated accounts or workstations, no just-in-time privilege elevation or multi-admin approvals, and overly permissive Graph API access. This article includes suggestions to remediate these issues, such as fixing the above issues, locking down script and policy deployment, hardening device enrollment, and implementing additional alerting on high-risk activities.
VE Commercial Vehicles' My Eicher telematics platform for Indian truck and bus fleets exposed unauthenticated internal APIs that listed hundreds of thousands of customers, users, vehicles, and OTP codes. By walking the API path, Eaton found open endpoints that returned user records, encrypted passwords, and a full OTP history since 2021, enabling account takeover by matching mobile numbers to OTPs or resetting passwords via API. This access allowed remote control and tracking of entire fleets and exposure of around 76,000 sensitive identity documents before the core vulnerability was fixed in November 2025 after repeated disclosure attempts.
Sponsored
The open-source security auditor that proves exploitability. Point it at any repo — every finding ships a verdict, a source-to-sink trace, and the CWE, while adversarial agents disprove the false alarms. The noise never reaches you. Works with any open model. See where it lands on the DVGA benchmark.
zeedumper connects to Kubernetes clusters using your kubeconfig files and dumps component z-pages by retrieving them through the API server proxy.
grokpatrol is a tool to detect whether the Grok Build CLI was installed on your machine and whether it collected and queued your git repositories for upload to xAI. If so, it will notify you of any secrets that were uploaded with them.
Abstract provides a streaming-first security operations platform that separates security data sources from storage targets, runs in-stream detections, routes data into multiple formats like OCSF and ECS, and aims to reduce SIEM storage costs.
Tachyon's committed canary .env files containing AWS, Anthropic, and Postgres credentials were made public on GitHub. GitHub flagged secrets within seconds without email alerts. AWS attached AWSCompromisedKeyQuarantineV3 shortly after a push, and within about five minutes, credentials connected externally. The quarantine policy didn't revoke keys or block secret commands, allowing an attacker using TruffleHog and Boto3 to access Secrets Manager and retrieve credentials seven minutes after pushing. Other traffic came from OVH, Tor exit, and Gigahost addresses, logging into Postgres and reading decoy tables. The intrusion stopped there, as the harvested secrets remained unused during the observation. Logs don't reveal if an LLM helped triage, leaving uncertainty about whether it was an attacker or defensive scan.
UK AISI and CAISI jointly evaluated Moonshot AI's Kimi K3, released July 16 and slated for open-weight release by July 27, and placed it well behind frontier US models while ahead of GLM-5.2, scoring 32% on Carnegie Mellon's ExploitBench against 76.2% for top US models and 24.4% for GLM-5.2, and reaching step 17 of the 32-step "The Last Ones" cyber range against 28.5 steps for the leading US models. The gap narrows unevenly because Kimi K3 matched the frontier on coverage and came close on bug reproduction but achieved arbitrary code execution on zero of 41 samples where top models managed 20, and yet it completed the full cyber range once in ten attempts, which the evaluators read as autonomous capability against small, weakly defended enterprise networks given initial access. Two framing points deserve weight, the first being that Kimi K3's safeguards did not stop it from attempting exploit development or offensive operations at all, and the second being that US closed-weight comparators ran with system-level safeguards disabled to measure maximum capability, so the published gap measures raw model ability rather than what a user encounters, a distinction that matters as open-weight releases from PRC labs keep closing the trendline.
A Claude Cowork session on macOS can chain a user-namespace feature, a Linux traffic-control module, and a recent kernel bug to escalate from an unprivileged VM user to guest root and then write directly to the host filesystem via a shared / mount. The write primitive comes from CVE-2026-46331 in act_pedit, and the path persists as similar bugs appear. To harden the environment, disable unprivileged user namespaces, tighten seccomp, block unneeded autoloaded modules and helper binaries, and scope or remove the host filesystem share so guest-root cannot touch SSH keys or other sensitive data.
GitHub paid researcher @sagitz_ a $100,000 bounty, reportedly among the largest in its Vulnerability Reward Program, for CVE-2026-3854, an unauthenticated remote code execution flaw in the handling of crafted repository URLs.
Dependabot now holds version update pull requests for 72 hours by default so a malicious package flagged minutes after publication has time to be pulled before it lands in a repo.
Sponsored
The "DarkSword" exploit chain, active since at least November 2025, chained six flaws across JavaScriptCore (CVE-2025-31277, CVE-2025-43529), dyld PAC bypass (CVE-2026-20700), WebContent sandbox escape (CVE-2025-14174), GPU sandbox escape (CVE-2025-43810), and a local privilege escalation (CVE-2025-43520) to achieve full kernel control on iOS 18.4 through 18.7 via a malicious Safari webpage. Multiple threat actors reused the same core chain across separate campaigns in Saudi Arabia, Turkey, Malaysia, and Ukraine, with delivery methods ranging from Snapchat-themed lures to compromised watering-hole sites, with PARS Defense linked to the Turkey and Malaysia activity. Apple patched all underlying flaws in stages through iOS 26.3 by February 11. Users should update immediately to close the fully patched chain.
Aura, an identity theft protection company, announced that one of its employees was targeted in a vishing attack. The attacker accessed 900,000 records, mostly containing only names and email addresses, with contact details like home addresses and phone numbers also accessed for up to 20,000 active customers and 15,000 former customers. Aura confirmed that no sensitive information such as SSNs or financial details was compromised.
A hacker going by the alias “THE INTERNET YIFF MACHINE” leaked more than 8.3M highly sensitive records from the tip and intelligence management company P3 Global Intel. The leak contains extensive personal data on those accused by tipsters, including names, email addresses, dates of birth, phone numbers, home addresses, license plate numbers, SSNs, and criminal histories. The hacker also disclosed that the company enables clients to collect a wealth of tracking data on “anonymous” tipsters.
Huntress documented a 277% surge in RMM abuse, with threat actors daisy-chaining tools like Action1 and ScreenConnect via MSI installers, wscript, and LLM-generated infostealer scripts to fragment telemetry, distribute persistence, and complicate attribution across campaigns targeting financial accounts and SSA-themed lures. Post-access tradecraft included pin.exe masquerading as Windows Security to harvest login PINs to ScreenConnect\Temp\output.txt, Sordum's HideUL.exe to conceal RMM installs from Add/Remove Programs, and WebBrowserPassView to harvest credentials, with C2 notifications routed through Telegram bots. Defenders should allowlist approved RMM tools explicitly, treat unrecognized RMM activity as suspicious by default, monitor for unexpected MSI installations from user-writable paths, and reference lolrmm.io for visibility into commonly abused platforms.
Eclypsium captured two previously undocumented malware variants on March 6: CondiBot, a Mirai-derived multi-architecture DDoS botnet (arm, mips, x86) with 32 registered attack handlers, competitive botnet killing, and C2 beaconing via port 20480 (0x5000) to 65.222.202.53; and Monaco, a Go 1.24.0 SSH scanner and XMRig-based Monero cryptominer attributed to a likely Chinese-speaking actor hosted on Alibaba Cloud (8.222.206.6), brute-forcing ~3.6 billion IPs with 50+ hardcoded credentials and reporting stolen creds back over raw TCP. Both variants exploit the EDR/XDR visibility gap on network appliances by operating below the OS layer. Defenders should monitor for /tmp/monaco, unauthorized chmod 777 operations, unexpected XMRig processes, and apply YARA rules from Eclypsium's full report to detect CondiBot artifacts including the "QTXBOT" string identifier.
AWS Security Agent allows customers to launch automated, agentic penetration tests on sites that they own. This post walks through the process of setting it up to scan DVWA running on an EC2 instance. The author was impressed by the presentation of the findings, such as including PoCs and verification, and felt that it can definitely augment penetration tests and reduce time to test, but felt that improvements are still needed, such as the ability to export results as a PDF.
Sponsored
The math on centralizing every log in your SIEM stopped working a while ago. In this webinar, Chronosphere, a Palo Alto Networks Company, and Google Cloud break down how to build a telemetry pipeline [11] that cuts ingestion costs without sacrificing detection coverage — covering filtering strategies, field reduction, and common
Pompelmi is an open-source Node.js upload security library that performs local, in-process scanning for spoofed files, archive bombs (ZIP traversal and nesting), polyglots, and script-bearing document structures without requiring a cloud API or daemon. It exposes typed verdicts with structured reasons for allow, quarantine, and reject flows, and ships framework adapters for Express, Next.js, NestJS, Koa, and Fastify with optional YARA and ClamAV integration.
ics-phishing-toolkit is a toolkit for remediating malicious calendar invites for teams using email solutions that don't natively remediate these issues.
tuxid is a lightweight, POSIX-compliant shell script that collects hardware, system, and network signals to generate a unique, reproducible fingerprint for a Linux machine.
OFAC sanctioned six individuals and two entities tied to the DPRK IT worker scheme (Coral Sleet/Jasper Sleet/Wagemole), which uses stolen identities, AI-generated personas, and Faceswap-altered documents to place North Korean operatives at Western companies, funneling salaries back to fund WMD programs. The operation runs through a multi-tiered structure of recruiters, facilitators, and western collaborators sourced via LinkedIn and GitHub, with operators tunneling traffic through Astrill VPN's US exit nodes from China-based infrastructure to masquerade as domestic employees. Post-access activity included proprietary data theft, extortion, and the use of agentic AI to generate and refine malware components. Microsoft has advised defenders to treat these intrusions as insider-risk scenarios and to monitor for abnormal credential use and low-and-slow access patterns.
FedRAMP spent nearly five years, 480 hours of review time, and 18 technical deep-dive sessions attempting to obtain basic data flow diagrams from Microsoft for its Government Community Cloud High (GCC High) offering, used by the Justice and Energy departments to protect information whose exposure "could be expected to have a severe or catastrophic adverse effect" on government operations. Microsoft's third-party assessors, Coalfire and Kratos, privately back-channeled to FedRAMP that it was "difficult to impossible" to obtain sufficient documentation, while a 2024 review team found fundamental issues with vulnerability remediation and scanning in Exchange Online and Teams alone, yet FedRAMP authorized GCC High on December 26, 2024, solely because widespread federal and defense-sector deployment made rejection impractical. DOGE has since gutted FedRAMP to a $10M annual budget with a skeleton staff, effectively reducing the program to a rubber stamp. At the same time, agencies are being pushed to adopt cloud-based AI tools handling reams of sensitive government data.
After a Meta employee posted a question on an internal forum, another engineer asked an AI agent to analyze the question, and the agent posted a response without the engineer's approval. The original poster took action on this advice and inadvertently exposed massive amounts of company and user data to engineers who were not authorized to have access to it. This follows a post last month by a safety and alignment director at Meta Superintelligence that an OpenClaw agent deleted her entire inbox despite being instructed to request confirmation before acting.
IBM X-Force and Flare Research mapped the full org chart of North Korea's fake IT worker operation, revealing 100,000+ workers across 40 countries earning $500M annually for Pyongyang via a structured hierarchy of recruiters, facilitators, and Western collaborators using tools like OConnect VPN and IP Messenger to secure and maintain remote IT roles at Western companies.
Eclypsium disclosed nine vulnerabilities across four low-cost KVM-over-IP devices from GL-iNet, Angeet/Yeeso, Sipeed, and JetKVM, including a CVSS 9.8 pre-auth RCE chain in the Angeet/Yeeso ES3 and unsigned firmware update mechanisms that could allow backdoored images.
Ubiquiti patched CVE-2026-22557 (CVSS 10.0), a path traversal flaw in UniFi Network Application v10.1.85 and earlier enabling account takeover, alongside CVE-2026-22558 (CVSS 7.7), an authenticated NoSQL injection enabling privilege escalation.
Sponsored
BloodHound Enterprise users can now see identity attack paths across Okta, GitHub, and Jamf using OpenGraph [4].
SentinelOne observed attackers exploiting three critical CVSS 9.8 vulnerabilities in FortiGate NGFWs between December 2025 and February 2026: CVE-2025-59718 and CVE-2025-59719, both stemming from improper cryptographic signature verification, which allowed unauthenticated attackers to forge SAML tokens and gain administrative access. CVE-2026-24858 was exploited as a zero-day to log into devices using an alternative account. CVE-2025-59718 was added to CISA's KEV catalog in late January 2026. Fortinet responded by suspending FortiCloud SSO before releasing a firmware patch. Defenders should apply the patch immediately, rotate all LDAP and AD credentials associated with FortiGate devices, enforce strong admin access controls, audit mS-DS-MachineAccountQuota settings, and monitor EDR telemetry from servers adjacent to the NGFW for unauthorized local admin account creation.
The pro-Iranian hacktivist group Handala compromised Stryker's Microsoft Intune environment on March 11, abusing legitimate MDM remote-wipe functionality to factory reset tens of thousands of Windows endpoints across 79 countries, erasing up to 95% of devices in some offices before containment. Handala, assessed by Palo Alto Networks to operate under the Iranian Ministry of Intelligence and Security, claims to have wiped over 200,000 systems and exfiltrated 50 TB of corporate data prior to triggering the destructive wipe. Medical devices, Vocera, Mako surgical robotics, LIFEPAK, and SurgiCount platforms remained unaffected due to architectural isolation from the impacted Microsoft environment. Organizations should audit MDM admin credential access and enforce conditional access policies to prevent similar living-off-the-land abuse.
Marquis had 672,075 people's data stolen in an August 2025 ransomware attack. Stolen data includes names, birth dates, addresses, bank and card account numbers, and Social Security numbers, with over half the victims living in Texas. Marquis sued its firewall vendor, SonicWall, in February, as it claims SonicWall's failings let attackers steal firewall configuration backup files to breach its network.
Semgrep Autofix, now in public beta, pairs the Semgrep Pro static analysis engine with frontier-model LLMs to deliver contextual upgrade guidance, line-level breaking change analysis, and AI-generated fix suggestions directly in pull requests. The tool performs dual static analysis — first-party code analysis to map how your codebase uses a dependency, and third-party version diffing to identify breaking changes — before passing results to an LLM to produce high-confidence remediation. Layered on top of Semgrep Assistant's existing 95% false positive reduction via codebase-aware reachability analysis, Autofix shifts developer effort from writing fixes to reviewing AI-generated patches.
JFrog launched the Agent Skills Registry, a centralized repository for governing AI agent skills — reusable, file-system-based instruction sets that execute with the invoking user's privileges and carry the same supply chain risks as OSS packages: prompt poisoning, malicious code, version drift, and weak provenance. The registry integrates with Agent Skills, ClawHub, and OpenShell, and enforces a publish-time security pipeline via `jf skills publish` that runs a two-stage behavioral scan, generates in-toto compliant attestations, and optionally produces cryptographically signed provenance evidence before a skill reaches the repository. At install time, `jf skill install` verifies that evidence chain, providing zero-trust consumption across coding agents, CI/CD pipelines, and automation tooling without locking teams into vendor-specific marketplaces.
watchTowr chained four flaws in BMC FootPrints — an auth bypass (CVE-2025-71257), two SSRF bugs (CVE-2025-71258, CVE-2025-71259), and a Java deserialization RCE (CVE-2025-71260) — to achieve pre-authenticated remote code execution on fully patched versions 20.20.02 through 20.24.01.001. The auth bypass leaks a session token via the password-reset endpoint, which then unlocks the SSRF and deserialization chains. BMC shipped hotfixes in September 2025 after a three-month back-and-forth reproduction process. CVEs were only assigned in March. FootPrints had no CVEs since 2014, making it an under-scrutinized target sitting on networks that also hold IT asset inventories and incident data.
Sponsored
Threat actors don't break in — they log in. As credential-based attacks become the primary entry point for breaches, you need workforce identity verification (IDV) to verify that every login is legitimate. The Workforce IDV checklist breaks down the seven must-have features to help you evaluate solutions and protect against unauthorized access.
VMkatz is a ~2.5 MB static binary that extracts Windows credentials directly from VM memory snapshots and virtual disks across VMware, VirtualBox, Proxmox, and Hyper-V without exfiltrating disk images. It supports all 9 LSASS SSP providers via in-place decryption, native VMFS-6 raw SCSI access to bypass file locks on running ESXi VMs, and natively parses NTDS.dit ESE databases for full Active Directory hash extraction with no impacket or external dependencies.
Native provides a multi-cloud security control plane that turns high-level security policy into provider-native controls, enforcing secure-by-design architecture across AWS, Azure, Google Cloud, and Oracle using built‑in security features instead of after-the-fact detection.
FIM is an Open Source Host-based file integrity monitoring tool that performs file system analysis, file integrity checking, real-time alerting, and provides Audit daemon data.
An OPSEC failure on an Iranian-hosted staging server at 185.221.239[.]162 exposed a financially motivated operator's full working environment, including a 15-node KCP-based relay network spanning Finnish Hetzner nodes and Iranian ISPs, a Python botnet deployer (ohhhh.py) opening 500 concurrent SSH sessions to compile and launch a DDoS bot client (cnc) directly on victim machines via gcc, and MHDDos tooling tested against live targets. The exposed .bash_history documented three operational phases: tunnel deployment using paqet and 3x-ui, DDoS development targeting a FiveM GTA server (5.42.223[.]60:30120) and 194.147.222[.]151, and iterative C2 botnet buildout with Farsi inline comments confirming operator origin. Defenders should block the listed IOCs, monitor for unexpected gcc invocations and renamed binaries ("hex"), audit SSH access logs for credential-stuffing patterns, and treat any recruited hosts as independently compromised regardless of C2 reachability.