Scattered Spider Suspect Grabbed at Helsinki Airport, Extradited to the US

A 19-year-old alleged Scattered Spider member was extradited from Finland to face US federal charges tied to more than 100 intrusions. The day was otherwise dominated by edge-appliance and Kubernetes RCEs, a massive Azure CLI password spray, and a fresh wave of offensive tooling and AI-agent attacks.

Threat Activity

• Peter Stokes, a 19-year-old US–Estonian dual citizen using the alias “Bouquet,” was extradited from Finland to a Chicago federal court after being detained at Helsinki Airport in April 2026 while trying to board a flight to Japan. The DOJ complaint links him to Scattered Spider activity spanning 100+ intrusions, ~$100M in ransom payments, and a 2025 breach of a “luxury-jewelry retailer.” The Record, DOJ.
• The FortiBleed credential-theft campaign has been linked to the INC and Lynx ransomware operations, suggesting the harvested Fortinet credentials are being staged for follow-on network intrusions. BleepingComputer.
• CSE (Canada) confirmed a Russian group breached a Quebec water-treatment plant, gaining access to control pumps and chlorine dosing systems — part of a surge in state-backed ICS targeting (3,200+ incidents in 2025–2026). Yahoo/CP.
• ChocoPoC, a Python RAT, is being distributed inside weaponized proof-of-concept exploits on GitHub in a campaign aimed at security researchers — a reminder to detonate untrusted PoCs only in isolation. BleepingComputer.
• Unit 42 detailed EtherRAT, delivered via Microsoft Teams phishing where an impersonated “System Administrator” seizes remote control and drops a malicious MSI/PowerShell payload — with its C2 configuration hidden inside an Ethereum smart contract. Unit 42.
• Kaspersky tracked a “massive, multi-domain” SEO-poisoning campaign abusing ScreenConnect to deploy AsyncRAT via fake installers for OBS Studio, DS4Windows, Bandicam and others. The Hacker News.
• TransparentTribe (APT36) is weaponizing the 2026 Indian NEET exam-leak controversy as a lure, deploying the Go-based StealthServer backdoor (Windows counterpart of DeskRAT) with WebSocket C2. FalconFeeds.

ClickFix & Delivery Techniques

• New research analyzing 3,000+ live ClickFix payloads shows the fake “prove you’re human” pages are now backed by API-driven servers that hand each visitor the same malware in a fresh disguise, plus a new delivery method built to evade Windows script scanning. BertJan Weesie’s companion write-up (from his OrangeCon session) argues ClickFix is now the dominant initial-access technique, not the exception. The Hacker News, kqlquery, Dark Reading.

Cloud & Identity

• Huntress is tracking an ongoing, automated password-spray campaign against Azure CLI, abusing the OAuth ROPC flow (which bypasses MFA) from an IPv6 range (2a0a:d683::/32) attributed to LSHIY LLC / AS32167 and linked to China. Over 81 million login attempts between June 12–26 compromised at least 78 Microsoft accounts across ~64–78 orgs. Recommended mitigations: enforce comprehensive MFA and block ROPC. Huntress, BleepingComputer.

Vulnerabilities & Exploits

• Progress Kemp LoadMaster pre-auth RCE CVE-2026-8037 (root command execution via crafted API request) is now under active exploitation, per eSentire’s TRU — moving from last week’s watchTowr PoC to real-world attacks. Patch if the API is exposed. The Hacker News.
• Synacktiv disclosed an unauthenticated RCE in Argo CD’s repo-server enabling full Kubernetes cluster takeover via malicious manifests and Redis cache abuse — reachable by anyone who can hit the internal port. There is no fix and no CVE yet; the write-up includes the CodeQL analysis used to find it. Synacktiv, The Hacker News.
• Cato AI Labs disclosed DuneSlide — two critical zero-click prompt-injection flaws in Cursor (CVE-2026-50548, CVE-2026-50549, both ~9.8) that let a single crafted prompt escape the editor sandbox and run arbitrary commands on a developer’s machine with no approval prompt. Fixed in Cursor 3.0. The Hacker News, Cato.
• Adobe patched seven CVSS 10.0 flaws across ColdFusion (arbitrary code execution, path traversal, SSRF, file read/write) and Campaign Classic — none under active attack. Notably, Adobe is shifting to twice-monthly patch cycles specifically because frontier AI models are now finding bugs at scale. BleepingComputer, Adobe.
• Over 900 Oracle E-Business Suite instances remain exposed online amid ongoing exploitation of the critical EBS flaw. BleepingComputer.
• Kaspersky flagged active exploitation of CVE-2024-2658 in Schneider Electric FlexNet Publisher — local privilege escalation via crafted openssl.cnf and DLL loading against ICS environments. Securelist.
• A new write-up details CVE-2026-31694, an out-of-bounds write in the Linux FUSE readdir cache yielding unprivileged-to-root privilege escalation. cyberstan.co.uk.
• Apple “Hide My Email” was found leaking users’ real addresses, undermining the masking service’s core privacy guarantee. EasyOptOuts.

AI & Model Security

• Palo Alto Unit 42 documented “Phantom Squatting”: attackers systematically probe LLMs to identify which non-existent domains models most often hallucinate for target brands (“adversarial hallucination probing”), then pre-register the highest-value ones and stand up phishing/malware within hours. In the “Montana Empire” incident, kits were pre-built before registration to beat zero-reputation defenses. Unit 42, Dark Reading.
• Check Point documented the first case of a frontier model (DeepSeek) being jailbroken into building working in-browser ransomware that abuses the browser File System Access API with AI-generated obfuscation, running entirely in-browser on Windows and Android. The Register, The Hacker News.
• Pentera Labs showed that hijacking a developer’s Claude Desktop account (via poisoned preferences and malicious prompts) can yield remote code execution, underscoring the need for sandboxing and privilege controls around agentic tools. The Register.
• Researchers found Claude Code (v2.1.196) was steganographically marking requests — encoding routing/telemetry info via subtle date and punctuation changes keyed to environment variables, timezone, and hostname. The Decoder reports the feature effectively flagged Chinese users; Anthropic is removing it after backlash. thereallo.dev, The Decoder.
• The US Commerce Department lifted export controls on Claude Fable 5 and Mythos 5, and Anthropic began restoring worldwide access after the two-week, jailbreak-linked ban; a new safety classifier reportedly blocks the technique in 99%+ of cases. The Record, The Hacker News.

New Tools & Releases

• CredSpy — a Rust/open-source tool for unauthenticated Entra ID user enumeration and auth-method discovery via the public GetCredentialType API, revealing whether target accounts use passkeys, certificate auth, or passwordless push. Useful for recon and identity-posture assessments. GitHub.
• win32k-callback-detouring — a Beacon Object File abusing the win32k.sys kernel callback mechanism for arbitrary code execution / EDR evasion. GitHub.
• CDP-Enable-BOF — an x64 BOF that enables the Chrome DevTools Protocol against chrome.exe/msedge.exe for session hijacking and post-exploitation, paired with the “Modern Session Hijacking by Living off the DevTools Protocol” talk. GitHub.
• watchTowr released a detection artifact generator for CVE-2026-8451 (Citrix NetScaler pre-auth memory overread) alongside its technical write-up. GitHub, watchTowr Labs.
• KeyHunter — a fast Rust tool for finding and verifying leaked API keys across public GitHub repos, covering 45+ providers and checking whether exposed secrets are still active. GitHub.

Data Breaches

• Medtronic is notifying customers affected by a ShinyHunters breach; the group also listed Ingram Content Group and Fluke Corporation (previously named by Clop in 2025) on its leak site. BleepingComputer.
• DHS confirmed hackers breached the Homeland Security Information Network (HSIN), the sensitive info-sharing platform used by federal, state, local and private-sector partners. BleepingComputer.
• Kubota North America disclosed that attackers had access to some of its network systems for over a month earlier this year. BleepingComputer.

Industry & Policy

• Microsoft is accelerating its post-quantum cryptography migration to 2029, with Azure CTO Mark Russinovich citing a shifted “risk horizon” from quantum R&D advances. The Hacker News.
• The Godot game engine will no longer accept AI-authored code contributions, citing an inability to trust that heavy AI users understand their own submissions. PC Gamer.