CTIPilot
AI-generated · no human review · verify critical claims against the linked source. how it works →
‹Thu · 21 May 2026›
All daily briefs →
Daily brief · UTC day

Thursday, 21 May 2026

8 verified findings from 1 run · 1 update to prior coverage · the settled record for this UTC day, in the classic brief order.

CriticalUpdate Drupal core sites on PostgreSQL to 11.3.10, 11.2.12, 11.1.10, 10.6.9, 10.5.10 or 10.4.10CVE-2026-9082 · exploited · corrected 21 Oct 16:06Z
Criticality
Kind
Topic
Region
TL;DR · the day in one read
  1. 01ESET: China-aligned Webworm targeted governments in Belgium, Italy, Serbia and Poland with Discord and OneDrive implants. ESET reported on 2026-05-20 that the China-aligned group Webworm targeted governmental organizations in Belgium, Italy, Serbia and Poland, and a South African university, in 2025 with two new backdoors: EchoCreep, written in Go, which takes commands through Discord, and GraphWorm, which uses Microsoft Graph and OneDrive for command and control. A compromised third-party Amazon S3 bucket held its proxy configurations and files taken from Italian and Spanish governmental entities; ESET could not find the entry point, but the group scanned victims' web servers and ran a public SquirrelMail exploit (CVE-2017-7692) against a Serbian webmail target. →
  2. 02Verizon's 2026 DBIR puts exploitation at 31% of breaches; only 26% of KEV-listed flaws were fully remediated. Verizon's 2026 Data Breach Investigations Report, published on 2026-05-19 and built on more than 31,000 incidents and 22,000 confirmed breaches from November 2024 to October 2025, finds vulnerability exploitation at 31% of initial access, up from 20%, ahead of credential abuse at 13%. Only 26% of KEV-listed vulnerabilities were fully remediated in 2025 (38% the year before), median time to full patching rose from 32 to 43 days, and breaches involving a third party rose 60% to 48% of all breaches. →
  3. 03PinTheft gives a local user root through an RDS zerocopy double-free, wherever the RDS kernel module is loaded. V12 Security released PinTheft and a working exploit on 2026-05-19, relayed to oss-security by Gentoo's Sam James: a double-free in the Linux kernel's RDS zerocopy send path, rds_message_zcopy_from_user(), that the exploit turns into a page-cache overwrite of a SUID-root binary through io_uring fixed buffers for a root shell. The bug was patched upstream earlier in May and the kernel fix was published as CVE-2026-43494 (CVSS 7.8) on 2026-05-21; exploitation needs the RDS modules, which V12 found loaded by default only on Arch Linux among common distributions. →

01Active threats, incidents & disclosures3 items

HIGHCVE-2017-7692updatedNATOB2

Webworm (China-aligned) turns to European government targets with the EchoCreep (Discord) and GraphWorm (Microsoft Graph and OneDrive) backdoors, ESET reports

ESET Research published its analysis of Webworm, a China-aligned group active since at least 2022 that it links to SixLittleMonkeys and FishMonger, on 2026-05-20: in 2025 the group moved from Asia to governmental organizations in Belgium, Italy, Serbia and Poland, compromised a South African university, and replaced its Trochilus and McRat backdoors with two new ones, EchoCreep and GraphWorm (ESET, 2026-05-20). ESET told The Hacker News that a link to Space Pirates is tenuous at best (The Hacker News, 2026-05-20). ESET could not find the entry point, but an operator server showed nuclei and dirsearch scans of 56 targets in countries including Spain, Hungary, Belgium, Nigeria, Czechia and Serbia, and a public exploit for the SquirrelMail post-authentication RCE CVE-2017-7692 run against a Serbian webmail target, which suggests the group held that victim's credentials (ESET, 2026-05-20).

EchoCreep and GraphWorm. EchoCreep, written in Go, sends all its traffic through Discord API endpoints, decodes commands with base64 and AES-CBC-128, runs them through cmd.exe and uploads files as attachments; it reports to per-victim channels, named after the victim's IP address or IP address and hostname, that the operators created before it first ran, and it runs under a custom scheduled task and carries a modified timestamp (ESET, 2026-05-20). ESET decrypted 433 messages going back to 2024-03-21 across four victim channels; the first real compromise came on 2025-04-09, followed by curl downloads (ESET, 2026-05-20). GraphWorm runs at every user logon through registry Run keys and talks only to OneDrive endpoints of Microsoft Graph, with a valid cloud account and API tokens, fetching jobs from and uploading results to a per-victim folder; it encrypts traffic with AES-256-CBC and base64, stages command output in a local file that it deletes after upload, and likely uses Graph's upload-session endpoint for large files (ESET, 2026-05-20).

Proxies, staging and stolen data. The proxy tools WormFrp, a modified frp that reads AES-encrypted configurations from a compromised third-party Amazon S3 bucket, ChainWorm, SmuxProxy and WormSocket build internal, external and multi-hop proxies through rented servers and VPS hosts, the group staged tools in a GitHub repository, and both the repository and the S3 bucket have been taken down (ESET, 2026-05-20). The S3 bucket held virtual-machine snapshots, one with the state of an Italian governmental machine, and from December 2025 to January 2026 an mRemoteNG connection-configuration file and a Visio infrastructure diagram taken from a Spanish governmental entity; in October 2025 the operators also uploaded SharpSecretsdump, a tool that mimics Impacket's secretsdump (ESET, 2026-05-20).

This trend continued in 2025, with the attacks we observed targeting governmental organizations in Belgium, Italy, Serbia, and Poland.

Upon EchoCreep’s first execution, it does not attempt to create a new channel, but sends a message saying Up Success to a channel that already exists

In the case of GraphWorm, which uses Microsoft Graph API for C&C communication, we discovered that it uses OneDrive endpoints exclusively, specifically to get new jobs and to upload victim information.

ESET Research (WeLiveSecurity) 2026-05-20

ESET told The Hacker News that Webworm's links to Space Pirates is tenuous at best, citing the use of open-source RATs and a lack of concrete evidence tying the two clusters.

The Hacker News 2026-05-20
Correctionrun 2026-10-21T1148Z-auditbodyclassificationcvesevidenceheadlinesourcessourcing_notesummarytechniquestitleverification

FishMonger (which The Hacker News also calls Aquatic Panda (The Hacker News, 2026-05-20)), SixLittleMonkeys and Space Pirates are not other names for Webworm: ESET links Webworm to SixLittleMonkeys and FishMonger as related China-aligned groups (ESET, 2026-05-20) and calls the Space Pirates link tenuous at best (The Hacker News, 2026-05-20). EchoCreep does not create its Discord channels: it reports to channels the operators made before it first ran, which suggests they knew their targets or had gathered the details after initial access (ESET, 2026-05-20). The S3 bucket was a compromised third-party bucket that has since been taken down, and SharpSecretsdump, which mimics Impacket's secretsdump rather than dumping LSASS memory, was only seen uploaded to it (ESET, 2026-05-20).

threat21 May 05:00Zsingle-sourceOpen finding →
ROUTINEupdatedNATOB2

B1ack's Stash carding marketplace releases about 4.6 million stolen payment card records for free

SOCRadar reported on 2026-05-18 that B1ack's Stash, a dark-web carding marketplace operating since at least 2023, suspended about 8 million stolen CVV2 records from its inventory because sellers had resold its cards in competing shops, and released about 4.6 million of them as a free download (SOCRadar, 2026-05-18). Each record holds the full card number, expiry date, CVV2, cardholder name, billing address, e-mail, phone number and IP address; SOCRadar assesses that the data likely comes from e-commerce skimming or phishing operations, and about 70% of the records belong to US cardholders, followed by Canada, the United Kingdom, France and Malaysia (SOCRadar, 2026-05-18). After SOCRadar filtered out expired and duplicate records, roughly 4.3 million appear fresh and usable, and the marketplace has used free releases as marketing before, one million cards in April 2024 and over four million records in February 2025 (Security Affairs, 2026-05-20).

making American cardholders by far the most heavily represented victims

The most immediate threat is card-not-present (CNP) fraud.

SOCRadar 2026-05-18

After filtering, roughly 4.3 million records appear to be fresh and potentially usable.

Security Affairs 2026-05-20
Improvementrun 2026-10-21T1148Z-auditbodyclassificationentitiesevent_dateevidenceheadlinesourcessourcing_notesummarytagstechniquestitleverificationpriority

The release was reported on 2026-05-18 and is part of about 8 million CVV2 records the marketplace suspended (SOCRadar, 2026-05-18); the estimate of roughly 4.3 million fresh records is Security Affairs' account of SOCRadar's filtering (Security Affairs, 2026-05-20). SOCRadar assesses only that the data likely comes from e-commerce skimming or phishing (SOCRadar, 2026-05-18); no source names WooCommerce or Magento storefronts or a root cause shared across the marketplace's releases.

incident21 May 05:00Zsingle-sourceOpen finding →
NOTABLECVE-2024-12802exploitedupdatedNATOB2

SonicWall Gen6 SSL-VPN: patched-firmware appliances still open to the CVE-2024-12802 MFA bypass without manual LDAP remediation, exploited February to March 2026 by a suspected access broker

SonicWall's SNWLID-2025-0001 describes CVE-2024-12802 (CWE-305, CVSS 6.5): when SSL-VPN authenticates against Microsoft Active Directory, UPN and SAM account names are handled separately, so MFA can be configured independently for each and an attacker can bypass it through the alternative account name (SonicWall SNWLID-2025-0001); in practice an attacker with valid credentials authenticates without the MFA step (BleepingComputer, 2026-05-20). On Gen6 the fixed firmware does not remove the vulnerable LDAP configuration, so six manual steps are needed: delete the LDAP configuration that uses userPrincipalName, remove cached LDAP users, remove the SSL VPN user domain, reboot, recreate the LDAP configuration without userPrincipalName and take a fresh backup (BleepingComputer, 2026-05-20); Gen7 firmware version 7.2.0-7015 and Gen8 firmware version 8.0.1-8017 build those steps in (SonicWall SNWLID-2025-0001). ReliaQuest responded to intrusions between February and March 2026 on Gen6 devices that looked patched and assesses with medium confidence that they are the first in-the-wild exploitation; the attacker brute-forced VPN accounts with an automated tool, in one case after only 13 attempts, and reached a file server within 30 minutes (ReliaQuest, 2026-05-19), connected to it over RDP with a shared local administrator password, and tried to deploy a Cobalt Strike beacon and a vulnerable driver to disable endpoint protection, both blocked by EDR (BleepingComputer, 2026-05-20). ReliaQuest describes the tools as consistent with actors in the ransomware ecosystem and the logouts and later re-entries with other accounts as initial access broker activity (ReliaQuest, 2026-05-19); the researchers could not attribute the attacks, though the activity resembled the Akira ransomware group's (Cybersecurity Dive, 2026-05-19).

Triage: where no CLI-based VPN client is authorized or typical, any sess="CLI" event is a high-fidelity indicator; where scripted VPN authentication is legitimate, ReliaQuest says to baseline it and tune on volume, source IP and timing (ReliaQuest, 2026-05-19).

SSL-VPN MFA Bypass in SonicWALL SSL-VPN can arise in specific cases due to the separate handling of UPN (User Principal Name) and SAM (Security Account Manager) account names when integrated with Microsoft Active Directory

SonicWall SNWLID-2025-0001 2025-01-07

The sess="CLI" session type indicates scripted or automated VPN authentication.

The voluntary logouts, combined with re-entry attempts days later using different accounts in some environments, are consistent with initial access broker (IAB) activity.

ReliaQuest 2026-05-19

Researchers could not provide direct attribution for the attacks, but the threat activity was consistent with ransomware activity from the Akira group.

Cybersecurity Dive 2026-05-19
Correctionrun 2026-10-21T1148Z-auditaffected_productsbodyclassificationcvesentitiesevent_dateevidenceheadlineregionssectorssourcessourcing_notesummarytagstechniquestitle

SonicWall scores CVE-2024-12802 CVSS 6.5 with a vector that requires a low-privileged account (PR:L), not 9.1 for an unauthenticated attacker; 9.1 is the assessment of CISA's Authorized Data Publisher (SonicWall SNWLID-2025-0001) (ReliaQuest, 2026-05-19), and the attacker needs valid credentials, which here were brute-forced. The attacks are not attributed: ReliaQuest describes a likely initial access broker using tools of the ransomware ecosystem, and the researchers could not attribute them, only compare them with Akira's 2025 activity (ReliaQuest, 2026-05-19) (Cybersecurity Dive, 2026-05-19).

The six remediation steps come from SonicWall's advisory as BleepingComputer lists them (BleepingComputer, 2026-05-20). The detection signal ReliaQuest found is the sess="CLI" session type with Event IDs 238 and 1080 and a CLI-to-GMS transition, not UPN-format login names (ReliaQuest, 2026-05-19).

threat21 May 05:00Zmulti-sourceOpen finding →
NOTABLECVE-2026-43494updatedNATOB2

PinTheft: Linux kernel RDS zerocopy double-free (CVE-2026-43494) turned into a root exploit through io_uring fixed buffers, public PoC; V12 found RDS loaded by default only on Arch Linux among the distributions it tested

V12 Security released PinTheft with a proof-of-concept exploit on 2026-05-19, and Gentoo's Sam James relayed it to the oss-security list the same day; Aaron Esau of the V12 team found the bug (oss-security, 2026-05-19). In the RDS zerocopy send path, rds_message_zcopy_from_user() pins user pages one at a time; when a later page faults, the error path drops the pages already pinned while the scatterlist entries stay live, so message cleanup drops them again and each failed send steals a reference from the first page (oss-security, 2026-05-19). The exploit registers an anonymous page as an io_uring fixed buffer, which gives it a FOLL_PIN bias of 1024 references, steals those references with failing RDS zerocopy sends, lets the page be reclaimed as page cache of a SUID-root binary and overwrites it through the stale fixed-buffer pointer with a small ELF payload, so running the binary drops into a root shell (oss-security, 2026-05-19). It needs the RDS modules, io_uring enabled, a readable SUID-root binary and x86_64 for the included payload (BleepingComputer, 2026-05-20). The bug was patched upstream earlier in May (BleepingComputer, 2026-05-20), and the kernel CNA published that fix, which resets op_nents in rds_message_zcopy_from_user(), as CVE-2026-43494 (CVSS 7.8) on 2026-05-21 (ENISA EUVD-2026-31267).

PinTheft is a Linux local privilege escalation exploit for an RDS

oss-security (Sam James, relaying V12 Security) 2026-05-19

Sadly, the RDS kernel module this requires is only default on Arch Linux among the common distributions we tested

Linux users on affected distros are advised to install the latest kernel updates as soon as possible.

BleepingComputer 2026-05-20
Improvementrun 2026-10-21T1148Z-auditaffected_productsbodyclassificationcvesevent_dateevidenceheadlinekindsourcessourcing_notesummarytechniquestitleverification

V12 found the RDS module (CONFIG_RDS with CONFIG_RDS_TCP) enabled by default only on Arch Linux among the common distributions it tested; it did not establish that the module is off on every other distribution (oss-security, 2026-05-19). Aaron Esau of V12 found the bug and V12 released the exploit, which Gentoo's Sam James relayed to oss-security (oss-security, 2026-05-19). The kernel fix was published as CVE-2026-43494 (CVSS 7.8) on 2026-05-21 (ENISA EUVD-2026-31267), and the mitigation for hosts that cannot take the fixed kernel yet is to unload and block the rds_tcp and rds modules (BleepingComputer, 2026-05-20).

vulnerability21 May 05:00Zsingle-sourceOpen finding →
NOTABLECVE-2026-7507 +5updatedNATOA2

Keycloak 26.6.2 fixes 16 security issues, among them an OIDC login session fixation (CVE-2026-7507) and a WebAuthn execute-actions token replay (CVE-2026-37982)

The Keycloak project released 26.6.2 on 2026-05-19 with fixes for 16 security issues carrying 18 CVE ids, several of them in bundled third-party components such as Bouncy Castle (Keycloak, 2026-05-19); BSI's WID-SEC-2026-1612, first released at 22:00 UTC that day, lists nine of the Keycloak CVEs and rates them high (BSI WID-SEC-2026-1612). CVE-2026-7507, which Red Hat calls a critical session fixation flaw although it scores CVSS 7.5, lets an unauthenticated attacker pre-create an authentication session and lure a victim to a crafted link: the /login-actions/restart endpoint lacks CSRF protection and cookie ownership checks, so single sign-on authenticates the victim and the attacker takes over the required-action form, which in default deployments reaches full takeover of the master-realm admin account (Red Hat CVE-2026-7507). CVE-2026-37982 (6.8) lets an attacker who intercepts an execute-actions e-mail link replay its token and register their own WebAuthn authenticator on the victim's account, where WebAuthn required actions are enabled (Red Hat CVE-2026-37982). CVE-2026-37979 (6.5) lets any confidential client with valid credentials read, through the token introspection endpoint, claims of lightweight access tokens meant for other resource servers (Red Hat CVE-2026-37979); CVE-2026-4630 (6.8) lets an authenticated client that knows the UUID of another resource server's resource in the same realm read, change or delete it through the Authorization Services Protection API (Red Hat CVE-2026-4630); CVE-2026-37978 (4.9) lets a low-privilege administrator with the view-clients role read other users' profile and role data through the evaluate-scopes Admin API (Red Hat CVE-2026-37978); and CVE-2026-6856 lets an authenticator that sends self-attestation bypass the acceptable-AAGUID allowlist during WebAuthn registration (Keycloak issue #48388). Red Hat fixed CVE-2026-7507 in Red Hat build of Keycloak 26.2.16 (RHSA-2026:19594) and it, CVE-2026-37982, CVE-2026-37979, CVE-2026-4630 and CVE-2026-37978 in 26.4.12 (RHSA-2026:19596).

This allows for full takeover of the master-realm admin account in default Keycloak deployments.

Red Hat CVE-2026-7507 2026-05-19

Exploitation requires WebAuthn required actions to be enabled and the email link to be compromised.

Red Hat CVE-2026-37982

By knowing or obtaining a resource's unique identifier (UUID) belonging to another Resource Server within the same realm, the client could bypass authorization checks.

Red Hat CVE-2026-4630
Correctionrun 2026-10-21T1148Z-auditactionsaffected_productsbodyclassificationcvesentitiesevent_dateevidenceheadlineregionssectorssourcessourcing_notesummarytagstechniquestitleverification

CVE-2026-4630 is not a cross-realm flaw: it lets an authenticated client reach another resource server's resources within the same realm (Red Hat CVE-2026-4630). CVE-2026-7507 works through the /login-actions/restart endpoint, with an attacker-created authentication session and a crafted link the victim follows, not through a crafted state parameter (Red Hat CVE-2026-7507), and CVE-2026-37982 needs an intercepted execute-actions e-mail link (Red Hat CVE-2026-37982). The highlighted CVEs carry Red Hat CVSS scores from 4.9 to 7.5 rather than none, CVE-2026-37979, CVE-2026-4630 and CVE-2026-37978 need an authenticated client or administrator, and the release fixes 16 issues with 18 CVE ids, nine of which BSI lists (Keycloak, 2026-05-19) (BSI WID-SEC-2026-1612). No source links the release to the Akira ransomware group.

vulnerability21 May 05:00Zsingle-sourceOpen finding →
NOTABLECVE-2026-45829updatedNATOB1

CVE-2026-45829: ChromaDB's Python FastAPI server runs an attacker's model code before checking authentication (CVSS 4.0 10.0), no fix in 1.5.9

HiddenLayer discovered CVE-2026-45829 in ChromaDB's Python FastAPI server and gave it the maximum severity score (BleepingComputer, 2026-05-19); Hadrian, which published a technical analysis the same day, gives CVSS 4.0 10.0 and repeats HiddenLayer's name for it, "ChromaToast", (Hadrian, 2026-05-19). The create-collection handler loads the request's embedding-function configuration before it checks authentication: with trust_remote_code set to true and model_name pointed at a HuggingFace repository or local path the attacker controls, the server downloads and instantiates the model and runs the attacker's Python inside the uvicorn server process (Hadrian, 2026-05-19), and only then rejects the request, with a 500 in HiddenLayer's description (BleepingComputer, 2026-05-19). Every Python FastAPI server from 1.0.0 is affected while the default Rust frontend does not use this code path, and there was no fixed version at the time of writing, 1.5.9 being the latest (Hadrian, 2026-05-19). HiddenLayer reports an initial acknowledgement on 2025-12-16 followed by silence across more than ten follow-ups (Hadrian, 2026-05-19), and its Shodan queries found roughly 73% of internet-exposed instances running a vulnerable version (BleepingComputer, 2026-05-19); Hadrian dates the report to 2025-11-28 (Hadrian, 2026-05-19), BleepingComputer to February 17 (BleepingComputer, 2026-05-19).

Triage: HuggingFace downloads started by your own ingestion code are normal; the same download from the Chroma server process that your ingestion code did not initiate is the signal (Hadrian, 2026-05-19).

Only the Python server is affected. The Rust server doesn't go through this code path.

There is no fixed version at time of writing (latest is 1.5.9).

Hadrian 2026-05-19

The server rejects the request, returns a 500, and the attacker's payload has already run.

BleepingComputer 2026-05-19
Correctionrun 2026-10-21T1148Z-auditactionsaffected_productsbodyclassificationcvesevidenceheadlinesectorssourcessourcing_notesummarytechniquestitle

No server-wide setting turns trust_remote_code off: the flag travels in the attacker's own create-collection request, inside the embedding-function arguments (Hadrian, 2026-05-19), and the sourced mitigations are the Rust frontend, keeping the Python server off public networks and restricting access to the API port (BleepingComputer, 2026-05-19). The rejection that follows the payload is a 500 in HiddenLayer's description (BleepingComputer, 2026-05-19) and in Hadrian's own proof of concept, although Hadrian's summary says 403 (Hadrian, 2026-05-19). HiddenLayer discovered and scored the flaw (BleepingComputer, 2026-05-19), and Hadrian published the technical analysis (Hadrian, 2026-05-19).

vulnerability21 May 05:00Zmulti-sourceOpen finding →
ROUTINECVE-2026-42822updatedNATOA2

CVE-2026-42822: improper authentication in Azure Local Disconnected Operations allows privilege escalation over the network (CVSS 10.0), fixed in ALDO 2604

Microsoft published CVE-2026-42822 on 2026-05-18: improper authentication (CWE-287) in Azure Local Disconnected Operations (ALDO) allows an unauthorized attacker to elevate privileges over a network, scored CVSS 10.0 with a scope change and rated Exploitation More Likely, though not exploited or publicly disclosed at release (MSRC, 2026-05-18). Microsoft says the most realistic scenario is a malicious or compromised insider who already has access to the internal environment and holds tenant or user identifiers, credentials or tokens; an external attacker would first need access to the internal network and valid identity context, and ALDO is designed to run disconnected and isolated (MSRC, 2026-05-18). The fix is ALDO version 2604 or later, delivered only as a full system update through the Azure portal, and ALDO is a restricted offering whose updates reach approved, allow-listed customers (MSRC, 2026-05-18).

Improper authentication in Azure Local Disconnected Operations allows an unauthorized attacker to elevate privileges over a network.

The most realistic exploitation scenario involves a malicious or compromised insider with existing access to the customer’s environment.

Updates are not available as standalone patches and must be applied as a full system update through the Azure portal.

MSRC CVE-2026-42822 2026-05-18
Correctionrun 2026-10-21T1148Z-auditactionsaffected_productsbodyclassificationcvesevidenceheadlineprioritysourcessourcing_notesummarytagstechniquestitle

In practice the flaw is not a no-credentials, no-foothold remote bug: Microsoft's realistic scenario is a malicious or compromised insider with existing access and identity material, and an external attacker would first need access to the internal network and valid identity context (MSRC, 2026-05-18). The advisory does not say that cloud-managed Azure was already protected while manual stacks were not: the affected product is ALDO, and its fix is version 2604 or later, applied as a full system update through the Azure portal. No source says ALDO is common in German or Swiss federal administrations; Microsoft describes it as a restricted offering for approved customers (MSRC, 2026-05-18).

vulnerability21 May 05:00Zsingle-sourceOpen finding →

03Research, reports & policy1 item

NOTABLEupdatedNATOB2

Verizon 2026 DBIR: vulnerability exploitation overtakes stolen credentials as the leading initial access vector for the first time in 19 years

Verizon published its 2026 Data Breach Investigations Report on 2026-05-19, built on more than 31,000 security incidents, over 22,000 of them confirmed breaches, at organizations in 145 countries (Verizon 2026 DBIR); the incidents date from November 2024 to October 2025 (Help Net Security, 2026-05-20). Exploitation of vulnerabilities is the most common initial access vector at 31% of breaches, up from 20%, and for the first time in the report's 19 years it has passed stolen credentials (Verizon 2026 DBIR) (Verizon, 2026-05-19); credential abuse fell from 22% to 13%, partly because Verizon now tracks pretexting as its own vector, and would have been 16% without that change (Verizon 2026 DBIR). Only 26% of KEV-listed vulnerabilities were fully remediated in 2025, against 38% the year before, and the median time to full patching grew from 32 to 43 days (Verizon 2026 DBIR) (Help Net Security, 2026-05-20). Breaches involving a third party rose 60% and now account for 48% of all breaches (Verizon, 2026-05-19), while only 23% of third-party organizations fully remediated their MFA issues; ransomware was present in 48% of breaches, up from 44%, and 69% of ransomware victims did not pay (Help Net Security, 2026-05-20).

This is the first time in 19 years that it has surpassed stolen credentials as the biggest point of entry.

Verizon (press release via GlobeNewswire) 2026-05-19

For the 2026 edition, that means incidents that happened between November 2024 and October 2025.

In the year before, that percentage reached 38, so this is a considerable drop, Verizon analysts noted.

Help Net Security 2026-05-20
Correctionrun 2026-10-21T1148Z-auditbodyclassificationdeep_divedeep_dive_categoryentitiesevent_dateevidenceheadlinekindprioritysectorssourcessourcing_notesummarytechniquestitleverification

The report covers more than 31,000 incidents and 22,000 confirmed breaches in 145 countries (Verizon 2026 DBIR) that date from November 2024 to October 2025 (Help Net Security, 2026-05-20), so its third-party growth reflects 2025 data and says nothing about campaigns active in May 2026 such as TeamPCP or Mini Shai-Hulud. The 26% is the share of KEV-listed vulnerabilities fully remediated by the organizations polled, not a share of organizations bound by BOD 22-01 missing deadlines, and 23% is the share of third-party organizations that fully remediated their MFA issues (Help Net Security, 2026-05-20). Exploitation rose from 20% to 31% in one year rather than over three cycles, and credential abuse fell from 22% to 13%, a drop Verizon partly ascribes to tracking pretexting separately (Verizon 2026 DBIR).

annual-report21 May 05:00Zsingle-sourceOpen finding →

04Updates to prior coverage1 item

CRITICALCVE-2026-9082exploitedupdatedNATOA1

CVE-2026-9082: anonymous SQL injection in Drupal core on PostgreSQL, Highly critical, exploited and CISA KEV-listed

First published 2026-05-20 · open finding →

Updaterun 2026-05-21-77cdc4cdcvesprioritysectorssourcestagsbody

Drupal patches "highly critical" pre-auth SQL injection (CVE-2026-9082) on PostgreSQL-backed installs of Drupal 8.9–11.3; the Security Team warned that "exploits might be developed within hours or days" of advisory release. EU/CH government portals and university CMSes are the primary exposed surface (Drupal Security Team, 2026-05-20; NCSC-CH, 2026-05-19).

Drupal published SA-CORE-2026-004 on 2026-05-20 with CVE-2026-9082: a vulnerability in core's database abstraction API allows an attacker to send specially crafted requests resulting in arbitrary SQL injection for sites using PostgreSQL, it can be exploited by anonymous users, and it "can lead to information disclosure, and in some cases privilege escalation, remote code execution, or other attacks" (Drupal SA-CORE-2026-004). The pre-announcement had scored it 20 of 25 and warned that "exploits might be developed within hours or days" (Drupal PSA-2026-05-18).

Affected versions run from 8.9.0 up to (not including) 10.4.10, 10.5.0 before 10.5.10, 10.6.0 before 10.6.9, 11.0.0 before 11.1.10, 11.2.0 before 11.2.12 and 11.3.0 before 11.3.10, and the fixed releases are 10.4.10, 10.5.10, 10.6.9, 11.1.10, 11.2.12 and 11.3.10; Drupal 9 and 8.9 sites get best-effort patch files to apply manually (Drupal SA-CORE-2026-004), and Drupal 7 is not affected (Drupal PSA-2026-05-18). Sites using the Drupal Steward web application firewall are already protected from known attack vectors but should upgrade in the near future (CSO Online, 2026-05-20). NCSC.ch had carried the pre-announcement on 2026-05-19 (NCSC.ch Security Hub).

Verification & coverage notes1 run

2026-05-21-77cdc4cd · Claude Opus 4.7 · 10 entries published

  • Out-of-window drops (primary source older than 36h window): Exim CVE-2026-45185 "Dead.Letter", XBOW disclosure 2026-05-12, NCSC-NL advisory 2026-05-15; S2 surfaced an exploitation-confirmation quote from NCSC-NL but its publication date is ambiguous, so the item is held to the next run with the underlying advisory carried forward; Fortinet CVE-2026-44277 (FortiAuthenticator) / CVE-2026-26083 (FortiSandbox), Fortinet PSIRT FG-IR-26-128 dated 2026-05-12, NCSC-CH advisory 2026-05-13, both outside the 36 h window; SAP May 2026 Security Patch Day (CVE-2026-34260, CVE-2026-34263); SAP Security Notes dated 2026-05-12, outside window; CrowdStrike 2026 Financial Services Threat Landscape Report; both cited URLs carry publication date 2026-05-14 (not 2026-05-20 as the sub-agent return initially asserted), which puts the report 6 days outside the 36 h window. Item dropped from § 3 by iteration-1 verification (consistent with the SAP / Fortinet treatment in this same list); finance-sector audience may pick up the synthesis directly from the CrowdStrike press release or CrowdStrike blog.
  • Already-covered drops: Microsoft Fox Tempest malware-signing-as-a-service disruption, covered in 2026-05-20 active-threats; S3 re-surfaced the Microsoft Threat Intelligence and Microsoft On the Issues posts with no material new development beyond yesterday's coverage. Huawei VRP zero-day → Luxembourg POST 2025 nationwide outage, covered in 2026-05-20 active-threats with the same The Record + Security Affairs sources S4 re-surfaced; no new technical specificity or CVE assignment in this run.
  • Long-running-campaign rule application: Microsoft Exchange CVE-2026-42897, re-surfaced by S2 with the same Microsoft MSRC + Microsoft Exchange Team + NCSC-CH + Help Net Security sources cited in the 2026-05-16 deep dive and the 2026-05-18 UPDATE. No new exploitation attribution, no new patch, no new victim class, per the long-running-campaign rule (≤1 consolidated UPDATE per week unless something critical changes) the item is not re-issued. Per PD-13 the imminent CISA KEV remediation deadline (2026-05-29, US-FCEB-only) is not a valid driver for a § 4 UPDATE.
  • Reduced-confidence / framing: B1ack's Stash 4.6M card dump included as a dark-web claim with explicit "not confirmed by issuing institutions" framing; SOCRadar and Security Affairs both analyse the actual dump, but per-issuer attribution is unverified. Huawei VRP / Luxembourg (not in this brief, but referenced in the dropped-list); confidence remains MEDIUM in the sub-agent finding due to absence of any Huawei PSIRT advisory after ~10 months, no CVE assigned and no technical advisory specificity.
  • Contradiction: Grafana TanStack timeline, Grafana's own post-mortem (Grafana Labs, 2026-05-19) cites detection of the TanStack compromise on 2026-05-11; BleepingComputer's reporting (BleepingComputer, 2026-05-20) cites 2026-05-01 for the malicious-package consumption event. The brief reports Grafana's date on the basis that Grafana is the primary disclosing party for its own incident; the 10-day discrepancy may reflect the difference between the malicious-package's pull-time on the CI/CD runner and the detection event on Grafana's security team's timeline.
  • Single-source items: none in this brief; every published item carries ≥2 independent reputable sources or qualifies under the PD-5 national-CERT carve-out with the CERT acting as primary disclosing party.
  • CVEs that did not clear § 2 inclusion gates (PD-2 § 2 gates: CISA KEV, ENISA EUVD exploited=true or CVSS≥9.0, vendor/researcher report of ITW exploitation, or pre-auth RCE on widely-deployed internet-exposed software with public PoC): none dropped from in-window candidates in this run; out-of-window CVEs listed above were not evaluated against gates.
  • Sub-agent self-identification: all four cti-research sub-agents (S1, S2, S3, S4) returned with **Model:** and **Timestamps:** lines; the AI-content notice and Generated by: line collapse to the single distinct model Claude Sonnet 4.6 since all four research roles reported that model.
  • Verification disposition (Phase 5.7): five iterations ran with model rotation per v2.47 (iter-1 Opus, iter-2 Sonnet alt, iter-3 Opus cold, iter-4 Sonnet alt, iter-5 Opus cold); cumulative findings across iterations remediated in-line. Iter-5 returned NEEDS_FIXES with truth=1 (the unsupported "approximately 20 % the previous year" prior-year DBIR baseline in § 5 Headline shift paragraph) and one advisory (Grafana scope wording precision drift), both remediated post-verdict before commit. The brief publishes at the v2.46 5-cap safety valve with verification_residual_count=1 recording the final-iteration verdict as the cap-breach signal for the Ops dashboard. The notable verification finding-cluster of this run was iter-3's surfacing of multiple sub-agent attribution-discipline regressions in the § 4 TeamPCP UPDATE (five FIRESCALE / 417k / SSM specifics misattributed to Wiz), all remediated by aligning the body to Wiz's actual published technical detail.
  • Coverage gaps: databreaches-net (Cloudflare 403, Wayback empty; now 5 consecutive run failures); inside-it-ch (403 persistent, 4-run failure); sophos-xops (HTTP 503 persistent, 4-run failure); trendmicro-research (HTTP 500 persistent, no Wayback snapshot); cyberscoop (TLS certificate not-yet-valid error); darkreading (HTTP 403, no Wayback); cert-fr-actu (feed stale since October 2025); us-treasury-ofac (503); cert-eu (200 empty-body, 2-run failure); edpb, cnil-fr, ico-uk, agid-csirt-it, quiet in window (no in-window enforcement items); ncsc-ch-security-hub bridge subcommand and chrome-releases and jpcert not fetched in this run.

Unmatched action items (migrated)

  • Patch Drupal core on PostgreSQL backends immediately: upgrade to 10.4.10 / 10.5.10 / 10.6.9 / 11.1.10 / 11.2.12 / 11.3.10. The Drupal Security Team warned that exploits "might be developed within hours or days" of the SA-CORE-2026-004 advisory (see § 4 UPDATE). If patch deployment is gated by change-control, temporarily front the site with Drupal Steward or an equivalent WAF rule covering SQL-injection vectors at the DB-API layer.
  • Complete the six-step SonicWall LDAP reconfiguration on every Gen6 SSL-VPN appliance per SonicWall KB kA1VN0000000RBd0AM; firmware-update status alone is insufficient and Akira-linked actors are actively exploiting the UPN/SAM split (see § 1). Given Gen6 EoL on 2026-04-16, schedule migration to Gen7/Gen8.
  • Hunt for Webworm Discord and Microsoft Graph API C2: alert on outbound HTTPS to discord.com/api/* or graph.microsoft.com from process trees whose parent is not the expected first-party application; correlate Graph API non-interactive sign-ins for app registrations without enterprise approval, and flag cmd.exe spawned by long-running services with no interactive user context (see § 1). Apply Conditional Access on Microsoft Graph restricting non-managed device sign-ins on workstations that have no Graph integration need.
  • Audit VS Code extension installation policies on every developer endpoint: enforce a managed allowlist via Group Policy / MDM, set extensions.autoUpdate: false, and rotate every secret accessible to extensions whenever a supply-chain compromise is confirmed (see § 4 UPDATE TeamPCP / GitHub breach). Hunt Sysmon EID 1 for code --install-extension invocations on dev endpoints; search CI/CD pipeline logs for durabletask package imports in versions 1.4.1–1.4.3 and treat any host that imported a malicious version as fully compromised.

Migrated from briefs/2026-05-21.md (v2).