Relay Chains, Bind-Link Blindspots, and a Wave of Live Zero-Days

A dense day for offensive tradecraft: SpecterOps documented a new NTLM-relay egress path for AD privilege escalation, Bitdefender showed how Windows bind links can blind EDR, and several new frameworks shipped. On the defensive front, two SonicWall SMA1000 zero-days are under active exploitation and CISA named three more SharePoint flaws being hit in the wild.

Offensive Tradecraft

• SpecterOps detailed “There and Back Again,” a technique to escalate Active Directory privileges by coercing outbound NTLM authentication through cloud relays over WebDAV — enabling identity takeover even when SMB egress is blocked. Mitigation centers on enforcing LDAP signing/channel binding and restricting outbound SMB. SpecterOps
• Bitdefender showed Windows bind links (bindflt.sys) can create conflicting filesystem views that hide malware from EDR. Redirecting a trusted path to attacker-controlled content leaves few artifacts, defeating path-based detection and application control. Bitdefender, SecurityWeek
• A PoC for ASUS’s bsitf.sys driver (CVE-2026-13585) grants arbitrary physical memory mapping via an unvalidated IOCTL — a classic BYOVD building block for kernel-level primitives. ahmadz.ai

New Tools & Releases

• ADPathFinder (NetSPI): unifies BloodHound CE, MSSQLHound, and ConfigManBearPig data via OpenGraph to map and prioritize shortest attack paths across AD, ADCS, SCCM, and MSSQL, with password auditing and escalation reports. GitHub, NetSPI blog
• KHAØS: a modern post-exploitation C2 framework added to the C2 Matrix, offering five covert channels and a full evasion feature set. c2_matrix (X)
• UnwindRaven: a Windows x64 offensive research framework that constructs fully synthetic call stacks at thread startup — relevant to both callstack-spoofing tradecraft and detection engineering. GitHub
• rsnap (Registry Snapshots): rapid post-exploitation Windows artifact enumeration by snapshotting the registry. Announcement
• deepteam: an open-source toolkit that simulates jailbreaking and prompt-injection attacks to surface vulnerabilities in LLM systems. GitHub
• k8s-aibom (Google Cloud): a read-only Kubernetes tool that generates AI/ML Bills of Materials to inventory and audit shadow AI workloads. SC World

Vulnerabilities & Exploits

• Two SonicWall SMA1000 zero-days are under active exploitation and now in CISA’s KEV catalog. CVE-2026-15409 (CVSS 10.0) is an unauthenticated SSRF, and CVE-2026-15410 can enable arbitrary command execution; patch immediately. Sweden’s national CERT has amplified the advisory. CERT-SE, The Hacker News, BleepingComputer
• CISA named three actively exploited on-prem SharePoint flaws — CVE-2026-32201, CVE-2026-45659, and CVE-2026-56164 — two of them hit as zero-days, affecting Subscription Edition, 2019, and 2016 (earlier coverage). CISA, BleepingComputer
• A pre-authentication RCE in the ServiceNow AI platform was disclosed via detailed sandbox-escape write-ups, with a patch now available. Slcyber research, follow-up
• Continuing this week’s record Patch Tuesday, a technical write-up landed for CVE-2026-58629, a double-free in dxgkrnl’s CreateAllocation rollback path enabling arbitrary/double frees via re-read user memory (earlier coverage). Separately, Microsoft paused the July update for some Intel-based Dell devices over stability issues. dxgkrnl analysis, The Register
• Detection content shipped for the LegacyHive Windows User Profile Service EoP PoC, with GossiTheDog publishing a hunting query; the anonymous researcher is reportedly on a streak of nine Patch-Tuesday-timed zero-days (earlier coverage). LegacyHive KQL, SecurityWeek
• Oversecured published 176 vulnerabilities in Samsung preinstalled Android apps. Repository

Supply Chain

• Four to five @asyncapi npm packages were compromised to deliver a multi-stage botnet loader with info-stealing/RAT capabilities. Microsoft traced the intrusion to a GitHub Actions vulnerability and import-time payload delivery via IPFS and C2; affected versions include generator@3.3.1, generator-helpers@1.1.1, and specs@6.11.2. Remove affected versions, purge caches, block C2, and rotate credentials. Microsoft, The Hacker News, BleepingComputer
• Nearly 300 fake GitHub repositories impersonate legitimate software and security projects to push infostealer malware. BleepingComputer
• ExporTheft: 11 “AI Chat Exporter” Chrome extensions exfiltrate full chat content on PDF export despite store listings claiming “No uploads to external servers.” malext.io
• Three malicious npm packages impersonate blockchain libraries (e.g., solana-key-utils@1.0.2), per Nextron. Nextron (X)

Threat Activity

• The World Leaks extortion group dumped ~19,000 files (14.3 GB) tied to India’s Kudankulam Nuclear Power Plant, reportedly including ventilation/cooling drawings, supplier and inspection records, and a control-room floor layout. Contractor Reliance Group confirmed a “partial breach” of data hosted by third-party provider Yotta; the leak does not appear to touch the Russian-built reactor cores. MarioNawfal (X), DailyDarkWeb (X)
• Iran abused long-known SS7 telecom vulnerabilities to geolocate U.S. military and contractor personnel in the Middle East in the run-up to and early days of the conflict, according to FT reporting on Mobile Surveillance Monitor research. Finland’s NCSC flagged the report. NCSC-FI/TechCrunch, SC World
• Ransomware crews are increasingly pre-staging backup destruction to remove recovery options and force payment. Sophos’s 2026 report puts median payments at $769K and average recovery costs at $1.7M, with over half of attacks succeeding. Mandiant (X), Sophos
• Lazarus has augmented its RAT with a dedicated keylogger (KKernel.exe) using a three-tier cascaded capture chain — Raw Input first, falling back to a global low-level keyboard hook, then async polling of all 256 virtual keys — logging at sentence granularity with foreground-window context. blackorbird (X)
• Kaspersky detailed OkoBot, a multi-stage framework active since April 2025 whose module injects seed-phrase phishing into legitimate Ledger and Trezor desktop apps; it uses PowerShell, SSH tunnels, and process injection. The Hacker News, Securelist
• Intel 471’s “Operation Fake KickOff” abuses fake recruiters and SaaS to harvest Google Workspace credentials with real-time MFA bypass since April 2025; defenses include phishing-resistant MFA and domain monitoring. Intel 471
• European police disrupted two large fraud operations: Spanish authorities took down a €140M investment-fraud and BEC ring (four arrests), and Dutch police dismantled a global crypto investment scam running ~24 call centers and 700+ fake “advisers” since 2021. Spain — BleepingComputer, Netherlands — The Record

AI & Model Security

• OpenAI unveiled GPT-Red, an internal automated red-teamer that finds prompt-injection vulnerabilities at scale via self-play — reportedly succeeding in ~84% of test scenarios versus ~13% for human teams, with results fed into hardening GPT-5.6. One practitioner cautioned that AI testing AI “should not become the only judge of its own” defenses. OpenAI, MIT Tech Review (discussion)
• Unit 42 analyzed TuxBot v3 “Evolution,” a modular IoT botnet framework showing signs of LLM-assisted development — functional DDoS core, but several exploit and fallback features broken by development bugs. Unit 42, The Hacker News
• PrismML’s Bonsai 27B compresses a 27B-parameter open reasoning model to under 4 GB to run on a phone, reportedly retaining ~90% of performance; Apple is said to be testing the compression tech. The Decoder (discussion)

Industry & Policy

• The White House launched “Gold Eagle,” an AI-driven clearinghouse for vulnerability detection, prioritization, and patching across industry, critical infrastructure, and government, stemming from the June 2 AI executive order. Denmark’s CERT has flagged it for European practitioners. The Record, SecurityWeek, CERT.dk