|
| 1 | +--- |
| 2 | +title: "Crow's Nest - 2026-05-08" |
| 3 | +date: 2026-05-08T00:00:00+00:00 |
| 4 | +draft: false |
| 5 | +tags: ["roundup"] |
| 6 | +--- |
| 7 | + |
| 8 | +A roundup of 38 items curated from across the security community. |
| 9 | + |
| 10 | +## 📰 News |
| 11 | + |
| 12 | +- [Defender for Endpoint: restrict response actions on high-value assets](https://learn.microsoft.com/en-us/defender-endpoint/restrict-response-actions-high-value-assets) by Sean Metcalf. |
| 13 | +> Defender for Endpoint adds a public preview to restrict live response actions on high-value assets. The control SOC analysts running scripts as SYSTEM on tier 0 boxes have been asking for. |
| 14 | +
|
| 15 | +- [AAD Graph activity logs land in Azure Monitor](https://learn.microsoft.com/en-us/azure/azure-monitor/reference/tables/aadgraphactivitylogs) by Alberto Verza. |
| 16 | +> Microsoft now ships AAD Graph activity logs through Azure Monitor. Defenders finally get the table they have been waiting on for hunting credential abuse against Entra. |
| 17 | +
|
| 18 | +- [UK AISI evaluates GPT-5.5 cyber capabilities](https://www.aisi.gov.uk/blog/our-evaluation-of-openais-gpt-5-5-cyber-capabilities) by DANΞ. |
| 19 | +> UK AISI evaluation of GPT-5.5 cyber capabilities. The model completed an end-to-end exploitation chain that AISI estimates would take a human expert ~20 hours. |
| 20 | +
|
| 21 | +- [Defender removes DigiCert AuthRoot registry entries](https://bugzilla.mozilla.org/show_bug.cgi?id=2033170) by Florian Roth ⚡️. |
| 22 | +> Microsoft Defender flagged DigiCert AuthRoot registry entries and removed them on a chunk of endpoints. Looks like a detection mistake rather than a trust-store response, but worth knowing if you saw the alerts. |
| 23 | +
|
| 24 | +- [Canvas breach disrupts schools and colleges nationwide](https://krebsonsecurity.com/2026/05/canvas-breach-disrupts-schools-colleges-nationwide) by BrianKrebs. |
| 25 | +> Ongoing data extortion attack against the Canvas LMS disrupted classes at thousands of US school districts and universities. Attackers defaced the login page with a ransom demand threatening to leak data on 275 million students and faculty across nearly 9,000 institutions. |
| 26 | +
|
| 27 | +- [OpenAI launches GPT-5.5 Cyber for defensive workflows](https://x.com/daveaitel/status/2052518939817545766) by Dave Aitel. |
| 28 | +> OpenAI ships GPT-5.5 with Trusted Access for Cyber, plus a Limited Preview of GPT-5.5-Cyber. Targeted at secure code review, vulnerability triage, detection engineering, malware analysis, and patch validation. |
| 29 | +
|
| 30 | +- [Dirty Frag: universal Linux LPE chaining xfrm-ESP and rxrpc](https://dirtyfrag.io/) by Dominic Chell 👻. |
| 31 | +> Universal Linux LPE chaining xfrm-ESP and rxrpc bugs. No race, no panic on failure, deterministic. Even with the copy.fail mitigation applied, every major distro is still vulnerable. |
| 32 | +
|
| 33 | +- [Chrome silently installs a 4 GB Gemini Nano model](https://awesomeagents.ai/news/chrome-gemini-nano-silent-install) by Simone Margaritelli. |
| 34 | +> Chrome silently installs a 4 GB Gemini Nano model file on user devices, no consent prompt, and re-downloads it if you delete it. Turns out the binary is a 6.3B INT4 build of Gemma 3 Nano with FP16 scales. |
| 35 | +
|
| 36 | +- [TAPOcalypse Now: exploiting TP-Link smart devices](https://labs.taszk.io/articles/post/tapocalypse) by Gergely Kalman. |
| 37 | +> Taszk Security Labs releases TAPOcalypse: exploiting TP-Link smart devices via LAN, browser, and the cloud account. Vendor fixed some, the rest stand at embargo expiry. |
| 38 | +
|
| 39 | +- [LAB52: 142-page EasterBunny report on APT29](https://lab52.io/blog/easterbunny) by hasherezade. |
| 40 | +> LAB52 publishes a 142-page open-access report on EasterBunny, advanced espionage artifacts attributed to APT29. Plenty for IR teams to feed into hunts. |
| 41 | +
|
| 42 | + |
| 43 | +## 📝 Techniques and Write-ups |
| 44 | + |
| 45 | +- [Bypassing Windows authentication reflection mitigations](https://www.synacktiv.com/en/publications/bypassing-windows-authentication-reflection-mitigations-for-system-shells-part-1) by Aurélien Chalot. |
| 46 | +> Synacktiv kicks off a series on bypassing the CVE-2025-33073 mitigations to pop SYSTEM shells via authentication reflection. Part one is up. |
| 47 | +
|
| 48 | +- [VS Code Dev Tunnels: the accidental C2 framework](https://specterops.io/blog/2026/05/06/dev-tunnels-the-accidental-c2) by N7WEra. |
| 49 | +> REST then WebSocket then SSH then MsgPack RPC. SpecterOps takes apart VS Code Dev Tunnels and finds a C2 framework underneath, with remote exec and file ops on top. |
| 50 | +
|
| 51 | +- [cPanel/WHM auth bypass CVE-2026-41940](https://labs.watchtowr.com/the-internet-is-falling-down-falling-down-falling-down-cpanel-whm-authentication-bypass-cve-2026-41940?123=) by SinSinology. |
| 52 | +> Auth bypass in cPanel/WHM tracked as CVE-2026-41940, full writeup from watchTowr. Also: [high-fidelity scanner from assetnote](https://slcyber.io/research-center/high-fidelity-check-for-the-cpanel-authentication-bypass-cve-2026-41940). |
| 53 | +
|
| 54 | +- [OpenAI on the origin of frontier-model jailbreaks](https://openai.com/index/where-the-goblins-came-from) by K̵i̵r̵k̵ ̵T̵r̵y̵c̵h̵e̵l̵. |
| 55 | +> OpenAI on the origin and shape of jailbreak attacks against frontier models. Useful framing if you are building or defending agentic systems. |
| 56 | +
|
| 57 | +- [KernelToUserInjector dodges ETW threat-intelligence sensors](https://github.com/winterknife/EVENSTAR/tree/master/KernelToUserInjector) by winterknife 🌻. |
| 58 | +> winterknife updates KernelToUserInjector to dodge ALLOCVM, WRITEVM, PROTECTVM, and QUEUEUSERAPC ETW threat-intelligence sensors, with sample EtwTi logs in the repo. |
| 59 | +
|
| 60 | +- [Codex on the wire: agent features as tradecraft](https://www.originhq.com/blog/codex-on-the-wire) by winterknife 🌻. |
| 61 | +> Origin on Codex on the Wire: agent features become tradecraft when they ship documented IPC for remote sessions. One bind flag turns a compromised endpoint into a remotely controlled agent. |
| 62 | +
|
| 63 | +- [DSCourier: WinGet COM API as a post-exploitation primitive](https://eclipsesec.com/posts/DSCourier) by Bad Sector Labs. |
| 64 | +> Novel post-exploitation via WinGet's COM API, executing through Microsoft-signed binaries. Also: [Octoberfest7's BOF POC](https://github.com/Octoberfest7/DSCourier_BOF). |
| 65 | +
|
| 66 | +- [Goodbye secure pool, hello KDP pool](https://windows-internals.com/goodbye-secure-pool-hello-kdp-pool) by chompie. |
| 67 | +> Yarden Shafir on KDP pool, the Windows kernel feature replacing the secure pool. Quick read on how it works and what the migration looks like. |
| 68 | +
|
| 69 | +- [ARP around and find out: hijacking GPO UNC paths](https://trustedsec.com/blog/arp-around-and-find-out-hijacking-gpo-unc-paths-for-code-execution-and-ntlm-relay?hss_channel=tw-403811306) by codewhisperer84. |
| 70 | +> TrustedSec on hijacking trusted Group Policy UNC paths for code execution and NTLM relay, no rogue GPO infrastructure or SYSVOL modifications required. |
| 71 | +
|
| 72 | +- [Kerberos with Titanis: walking the auth flow](https://trustedsec.com/blog/kerberos-with-titanis?hss_channel=tw-403811306) by codewhisperer84. |
| 73 | +> TrustedSec walks the full Kerberos authentication flow and shows how to drive every stage of it with the Titanis toolset. |
| 74 | +
|
| 75 | +- [Two command injections in Windows context menus](https://specterops.io/blog/2026/05/07/shift-happens-uncovering-two-built-in-command-injections-in-windows-context-menus) by Remi GASCOU. |
| 76 | +> SpecterOps finds two command injection bugs in the Windows Explorer "Open PowerShell window here" context menu. A folder name like `folder; calc` triggers arbitrary PowerShell when the user shift-right-clicks. |
| 77 | +
|
| 78 | +- [MCP server OAuth registration is wide open](https://openyoutu.be/wyxveABHu3I) by Arun. |
| 79 | +> Most MCP servers ship with a wide-open OAuth registration endpoint. One curl command is enough to land an account takeover. Timely if you have AI agents wired into LAN-side MCP servers. |
| 80 | +
|
| 81 | +- [EDR internals research and bypass](https://0xdbgman.github.io/posts/edr-internals-research-and-bypass) by Arun. |
| 82 | +> A tour through EDR internals and the bypass primitives that fall out of them. The author flags it as their last blog post on the topic. |
| 83 | +
|
| 84 | +- [SpecterOps: what red team actually means](https://ghst.ly/4uk1qaj) by Arun. |
| 85 | +> SpecterOps on what red team actually means once you strip out the vendor-speak. The right question is whether you would detect an attacker already inside. |
| 86 | +
|
| 87 | +- [The defensive stack is exposed](https://trustedsec.com/blog/the-defensive-stack-is-exposed) by d3d aka dead (dead, мёртв, 死了). |
| 88 | +> AI is shifting the economics of probing the defensive stack itself. TrustedSec walks through how the tools defenders depend on are quietly becoming part of the attack surface. |
| 89 | +
|
| 90 | +- [dMSA Ouroboros: self-sustaining credential extraction on Server 2025](https://www.huntress.com/blog/dmsa-ouroboros-credential-extraction-windows-server-2025) by Andrea P. |
| 91 | +> Huntress on dMSA Ouroboros: six commands of self-sustaining credential extraction on patched Server 2025. Survives deletion of the original attacker account. |
| 92 | +
|
| 93 | +- [Trail of Bits beats Google's zero-knowledge proof of quantum cryptanalysis](https://blog.trailofbits.com/2026/04/17/we-beat-googles-zero-knowledge-proof-of-quantum-cryptanalysis) by Dan Guido. |
| 94 | +> Trail of Bits cryptanalyzes Google's zero-knowledge proof of quantum cryptanalysis and finds a way to forge it. Worth reading whether or not you follow PQC arguments. |
| 95 | +
|
| 96 | +- [Linux LPE wave: copy.fail across major distros](https://xint.io/blog/copy-fail-linux-distributions) by Gergely Kalman. |
| 97 | +> Trivially exploitable Linux LPE in coreutils, reachable on every major distro from the last nine years. Also: [meta site](https://copy.fail/), [retr0 explainer](https://retr0.zip/blog/cve-2026-31431-copy-fail.html), [Go port from Bad Sector Labs](https://github.com/badsectorlabs/copyfail-go), [Ubuntu Rust coreutils race conditions](https://seclists.org/oss-sec/2026/q2/332). |
| 98 | +
|
| 99 | +- [MAD Bugs: 21-year-old PHP unserialize use-after-free](https://open.substack.com/pub/calif/p/mad-bugs-finding-and-exploiting-a?r=26yra9&%3Butm_campaign=post&%3Butm_medium=web) by Stefan Esser. |
| 100 | +> A 21-year-old PHP unserialize use-after-free that survived two decades of audit. The Calif team welcomes Stefan Esser back with a fresh writeup. |
| 101 | +
|
| 102 | + |
| 103 | +## 🛠️ Tools and Exploits |
| 104 | + |
| 105 | +- [BlackArrow opens up its talks repository](https://github.com/blackarrowsec/talks) by BlackArrow. |
| 106 | +> BlackArrow opens up the materials from every talk they have given. Slides, PoCs, and code in one repo. |
| 107 | +
|
| 108 | +- [Invoke-ADLabBuildOut adds AD sites and subnets](https://github.com/PyroTek3/ADLab) by Sean Metcalf. |
| 109 | +> Sean Metcalf updates Invoke-ADLabBuildOut to spin up AD sites, subnets, and site links on top of the lab build, with common security misconfigurations baked in. |
| 110 | +
|
| 111 | +- [MSSQLHound rewritten in Go](https://github.com/Mayyhem/MSSQLHound) by S3cur3Th1sSh1t. |
| 112 | +> MSSQLHound rewritten in Go: 17 minutes down to 17 seconds in lab, plus SOCKS proxying, Kerberos and NT hash auth, and pathfinding into BloodHound. |
| 113 | +
|
| 114 | +- [SOAPy gains Kerberos auth over ADWS](https://github.com/logangoins/SOAPy) by Logan Goins. |
| 115 | +> SOAPy now does kerberos auth over ADWS in Python, an integration nobody had bothered to ship in two years. Bigger release coming. |
| 116 | +
|
| 117 | +- [net_use BOF: drive mapping without net.exe](https://github.com/atomiczsec/Adrenaline/tree/main/community/net_use) by Adam Chester 🏴☠️. |
| 118 | +> net_use BOF: a modernized port of TrustedSec's SA repo entry for adding, listing, and removing mapped drives via the Windows MPR API. Avoids the telemetry from spawning net.exe or PowerShell. |
| 119 | +
|
| 120 | +- [mona v3 released](https://github.com/corelan/mona3) by ς๏гєɭคภς0๔3г (corelanc0d3r@infosec.exchange). |
| 121 | +> mona v3 ships: Python 2 and 3, 32 and 64-bit targets, WinDBG and WinDBGX, faster and leaner. The exploit-dev sidekick gets a long-overdue refresh. |
| 122 | +
|
| 123 | +- [claude-red: offensive-security skills for Claude](https://github.com/SnailSploit/Claude-Red) by Jason Lang. |
| 124 | +> claude-red is a curated library of offensive-security skills for the Claude skills system, each one a structured SKILL.md priming Claude with methodology for a specific attack surface. |
| 125 | +
|
| 126 | +- [Puzzle: malware via Windows minifilter abuse](https://github.com/Kudaes/Puzzle) by hasherezade. |
| 127 | +> Puzzle deploys malware in monitored environments by abusing Windows minifilter functionality. Includes utilities and PoCs to interact with minifilters and explore static and runtime analysis evasion. |
| 128 | +
|
| 129 | +- [Microsoft EventLogExpert: an improved Event Viewer](https://github.com/microsoft/EventLogExpert) by Chris Nickerson. |
| 130 | +> Microsoft quietly publishes EventLogExpert, an improved Event Viewer for IT and helpdesk teams. Open source, modern UI. |
| 131 | +
|
| 132 | + |
| 133 | + |
0 commit comments