7 verified findings from 1 run · 1 update to prior coverage · the settled record for this UTC day, in the classic brief order.
Criticality
Kind
Topic
Region
TL;DR · the day in one read
01A rogue Checkmarx Jenkins AST plugin build that hunts credentials was live for about 31 hours; TeamPCP claimed it. A modified build 2026.5.09 of the Checkmarx Jenkins AST plugin, published to the Jenkins Marketplace outside the plugin's release pipeline, was live from 2026-05-09 01:25 to 2026-05-10 08:47 UTC; its payload walks an OS-specific list of file paths looking for credentials of crypto wallets, VPNs, AWS and GitHub (Checkmarx). Checkmarx says the attacker reused access obtained in the March 2026 Trivy compromise, and TeamPCP claimed the attack, the third wave of malicious Checkmarx artifacts since late March. Checkmarx recommends rotating every credential a pipeline that ran the build could reach. →
02A PAN-OS Captive Portal overflow gives unauthenticated attackers root on exposed firewalls; Unit 42 saw tunnelling and AD enumeration afterwards. CVE-2026-0300 (CVSS 4.0 9.3, 8.7 when the portal is restricted to trusted addresses) is a buffer overflow in the PAN-OS User-ID Authentication Portal that gives unauthenticated attackers root on PA-Series and VM-Series firewalls. Unit 42 ties the limited exploitation since April to one likely state-sponsored cluster, CL-STA-1132; CISA lists it as exploited, and fixed builds now exist for every supported branch. →
03GTIG finds a criminal zero-day exploit it believes was built with AI; its discovery may have prevented mass exploitation. Google Threat Intelligence Group's AI Threat Tracker of 2026-05-11 reports the first zero-day exploit it believes was developed with AI: a Python 2FA bypass for a popular open-source web-based system administration tool that needs valid credentials, which criminals planned to use in mass exploitation before GTIG's coordinated disclosure may have prevented it. The report also documents LLM-generated decoy code in the Russia-nexus CANFAIL and LONGSTREAM malware, the Android backdoor PROMPTSPY driving devices through Gemini, PRC and DPRK actors using Gemini for vulnerability research, pooled LLM access, and TeamPCP's supply-chain compromises of AI components. →
Škoda Auto Deutschland's security monitoring found that unauthorised parties had exploited a vulnerability in the standard shop software of its online accessories shop and gained temporary access to the shop system; the shop was taken offline, the vulnerability fixed, an external IT forensics firm engaged and the incident reported to the competent data-protection authority (Škoda Auto Deutschland). The shop stores names and addresses, email addresses and sometimes telephone numbers, order information and customer-account credentials made up of the email address and a password kept as a cryptographic hash; full credit card data is processed only by the payment providers (Škoda Auto Deutschland). Škoda's analysis found that access to the stored data was possible in principle, but the nature of its logs does not allow it to reconstruct whether and to what extent data was actually copied or retrieved, and it has no concrete indication of misuse (Škoda Auto Deutschland). Only the accessories shop run by Škoda Auto Deutschland GmbH for customers in Germany was affected, not other Škoda online shops, the systems of the manufacturer Škoda Auto a.s. or the Škoda Connect portal (Škoda Auto Deutschland). SecurityWeek reported the disclosure on 2026-05-11 and noted that the number of affected customers had not been disclosed (SecurityWeek, 2026-05-11).
Our technical security monitoring established that unauthorised parties had exploited a vulnerability in the standard shop software in use. (translated from German)
Because of the nature of the available logs, however, it cannot be reconstructed in every detail after the fact whether and to what extent data was actually copied or retrieved. (translated from German)
Access credentials consist of the email address and a password that is stored in encrypted form, that is, as a cryptographic hash. (translated from German)
The Škoda Connect portal is not affected by the security incident. (translated from German)
Škoda Auto Deutschland
The company has yet to disclose how many individuals were potentially affected.
Škoda Auto Deutschland reported the incident to the competent data-protection authority but does not say when, so no 72-hour timing can be stated (Škoda Auto Deutschland). Its own words on the data are that access to the stored data was possible in principle and that, because of the nature of the available logs, it cannot reconstruct in every detail whether and to what extent data was actually copied or retrieved (translated from German); the shop concerned is only the German accessories shop, while other Škoda shops, the manufacturer's systems and Škoda Connect are unaffected and full card data is not stored there (Škoda Auto Deutschland). The step Škoda gives customers is to change passwords reused from the shop and to watch for messages that refer to their Škoda orders (Škoda Auto Deutschland).
incident12 May 05:00Zsingle-source · victim disclosureOpen finding →
The UK Information Commissioner's Office fined South Staffordshire Plc and South Staffordshire Water Plc £963,900 for infringing Articles 5(1)(f) and 32(1) of the UK GDPR (ICO enforcement action), after a 40% reduction for the company's early admission, which it accepted without appeal (ICO, 2026-05-11). The attack began in September 2020 with a phishing email whose attachment, once opened, let the attacker install malicious software that stayed undetected for 20 months; in May 2022 the attacker moved through the network and compromised domain administrator privileges, and the breach was found only when IT performance problems started an internal investigation on 15 July 2022, followed by a ransom note on 26 July (ICO, 2026-05-11). Two domain controllers were unpatched against ZeroLogon (CVE-2020-1472) in May 2022, and while the incident did not start through it, the attacker exploited it during the incident (ICO penalty notice). The company held data on about 1.85 million customers, around 750,000 current and 1.1 million former, and on its staff; the personal data of 633,887 people was published on the dark web, part of over 4.1 TB of data, including online-service usernames and passwords, bank account numbers and sort codes, employees' National Insurance numbers and information from which disabilities of Priority Services Register customers could be inferred (ICO, 2026-05-11). The ICO does not name the attacker; The Record and The Register attribute the attack to Cl0p, which claimed it, and the penalty notice makes no reference to any compromise of operational or water-treatment systems (The Record, 2026-05-11; The Register, 2026-05-11).
In this case, the recipient opened an attachment which enabled the attacker to install malicious software which remained undetected within the organisation's systems for 20 months.
Inadequate monitoring and logging; only 5% of the IT environment was being monitored, meaning malicious activity was not detected.
Two domain controllers also remained unpatched against a critical vulnerability known as ZeroLogon which allows rapid escalation of privileges and was first published in August 2020. The attacker successfully exploited this vulnerability during the incident.
The penalty notice makes no reference to any compromise of operational or water treatment systems.
The ICO itself names the vulnerability: two domain controllers were unpatched against ZeroLogon (CVE-2020-1472) in May 2022, and the attacker exploited it during the incident, although the incident did not start through it (ICO penalty notice); it is the Cl0p attribution that comes from The Record and The Register (The Record, 2026-05-11). The sequence was a phishing attachment opened in September 2020, malicious software undetected for 20 months, then movement through the network and domain-admin compromise from May 2022, with the breach found on 15 July 2022 (ICO, 2026-05-11). The monitoring gap was concrete: a third-party security operations centre covered 5% of the IT environment and endpoint telemetry was not in the SIEM (ICO penalty notice). The fine is a UK GDPR penalty under Articles 5(1)(f) and 32(1) (ICO enforcement action); the Cyber Security and Resilience Bill was still expected to be introduced to Parliament (The Record, 2026-05-11).
A special unit of the Spanish National Police arrested the suspected 35-year-old German operator of the relaunched Crimenetwork marketplace at his residence in Mallorca under a European Arrest Warrant, in an operation with Germany's Federal Criminal Police Office (BKA) and the Frankfurt am Main Public Prosecutor's Office's Central Office for Combating Internet Crime (ZIT); he is accused under Section 127 of the German Criminal Code of operating criminal trading platforms, and of narcotics trafficking (Help Net Security, 2026-05-11). Authorities say he built new infrastructure for Crimenetwork a few days after the previous version was shut down and its administrator arrested in December 2024; the BKA calls the marketplace one of the central marketplaces of the German-speaking underground economy for many years (Help Net Security, 2026-05-11). The relaunched platform offered stolen data, drugs, forged documents and other illegal services to more than 22,000 users and over 100 vendors, took payment in Bitcoin, Litecoin and Monero and earned an estimated more than €3.6 million in commissions and vendor fees; about €194,000 in assets were preliminarily seized (Help Net Security, 2026-05-11).
A special unit of the Spanish National Police arrested the suspected 35-year-old German operator at his residence in Mallorca under a European Arrest Warrant.
Authorities state the suspect allegedly created and operated a new technical infrastructure for Crimenetwork a few days after the previous version was shut down and its administrator was arrested in December 2024.
User and transaction data obtained during the operation are expected to support further investigations into the platform’s users and operators, who are believed to be mainly located in German-speaking countries.
The arrest in Mallorca was made by a special unit of the Spanish National Police in the operation with the BKA and ZIT, and the BKA describes Crimenetwork as one of the central marketplaces of the German-speaking underground economy, relaunched a few days after its December 2024 takedown (Help Net Security, 2026-05-11). The seized user and transaction data are expected to support further investigations into the platform's users and operators, believed to be mainly in German-speaking countries (Help Net Security, 2026-05-11).
On Saturday 2026-05-09 a rogue build, version 2026.5.09, of the Checkmarx Jenkins AST plugin was uploaded to repo.jenkins-ci.org outside the plugin's release pipeline; it broke the plugin's version scheme and had no git tag or GitHub release (BleepingComputer, 2026-05-11). Checkmarx says an external service account modified the plugin and published it to the Jenkins Marketplace, that the malicious window ran from 2026-05-09 01:25 UTC to 2026-05-10 08:47 UTC, and that the attacker used access obtained in the March incident, which began with credentials taken in the Trivy supply-chain attack on 2026-03-19 (Checkmarx, updated 2026-07-06). It is the third wave of malicious Checkmarx artifacts after GitHub Actions and Open VSX extensions on 2026-03-23 and a KICS Docker image, VS Code and DevAssist extensions and a GitHub Action on 2026-04-22 (Checkmarx, updated 2026-07-06). The payload determines the operating system (Windows, Linux or macOS), then walks a list of file paths for common applications on that system looking for credentials, with crypto wallets, VPNs, AWS and GitHub among the targets; Checkmarx had found no references to Jenkins in it, and Mandiant found no threat-actor access to the Checkmarx One SaaS environment (Checkmarx, updated 2026-07-06).
TeamPCP claimed the compromise (BleepingComputer, 2026-05-11), and researcher Adnan Khan and SOCRadar report that the group gained access to the plugin's GitHub repository and renamed it "Checkmarx-Fully-Hacked-by-TeamPCP-and-Their-Customers-Should-Cancel-Now", with the description "Checkmarx fails to rotate secrets again. with love – TeamPCP" (The Hacker News, 2026-05-11). Google Threat Intelligence Group tracks TeamPCP as UNC6780 and says that in the late-March compromises (Trivy, Checkmarx, LiteLLM, BerriAI) the group embedded the SANDCLOCK credential stealer to extract cloud secrets such as AWS keys and GitHub tokens from build environments (GTIG, 2026-05-11).
The malicious window was 2026-05-09 01:25 UTC to 2026-05-10 08:47 UTC.
Our assessment based on currently available evidence is that the threat actor was able to leverage access, obtained as part of the March incident, to later publish the modified version of the Jenkins plugin on May 9.
Targeted applications include crypto wallets, VPNs, AWS, and Github.
No CVE covers the rogue plugin build: CVE-2026-33634 is the Trivy flaw in which a threat actor used compromised credentials on 2026-03-19 to publish a malicious Trivy release and move the trivy-action tags to credential-stealing code (GitHub Advisory Database, 2026-03-21), the attack Checkmarx names as the source of the attacker's access (Checkmarx, updated 2026-07-06). No source names SANDCLOCK as the Jenkins payload; GTIG uses the name for the stealer of the late-March compromises and describes it extracting cloud secrets such as AWS keys and GitHub tokens, not LLM API keys (GTIG, 2026-05-11). Checkmarx lists crypto wallets, VPNs, AWS and GitHub among the applications the payload targets, and Mandiant found no threat-actor access to the Checkmarx One SaaS environment (Checkmarx, updated 2026-07-06).
The earlier waves were GitHub Actions and Open VSX extensions in March and the KICS Docker image with updated extensions and a GitHub Action in April (Checkmarx, updated 2026-07-06). PCPJack is a separate cloud worm whose first act is to evict TeamPCP tooling (SentinelLABS, 2026-05-07), and no cited source links it to the plugin. The defacement renamed the plugin's GitHub repository, not the build (The Hacker News, 2026-05-11), and exposure did not depend on auto-update: any instance that pulled build 2026.5.09 during the window installed the compromised plugin (SOCRadar, 2026-05-11).
West Pharmaceutical Services determined on 2026-05-07 that it had suffered a material cybersecurity attack in which an unauthorized party exfiltrated data and encrypted some systems; after detecting the intrusion on 2026-05-04 it took systems offline globally, notified law enforcement and engaged external forensic experts, and the 8-K filed on 2026-05-11 says the business was disrupted globally, core enterprise systems were restored and shipping, receiving and manufacturing had restarted at some sites, with no timeline yet for complete restoration and no access vector or exfiltration method stated (SEC 8-K, 2026-05-11). The company, which specialises in injectable drug packaging, syringe and vial components, containment systems and drug delivery devices, engaged Palo Alto Networks' Unit 42 for incident response, and no ransomware group had taken credit as of 2026-05-13 (BleepingComputer, 2026-05-13).
On May 7, 2026, West Pharmaceutical Services, Inc. (the “Company”) determined that the Company has experienced a material cybersecurity attack, in which certain data was exfiltrated by an unauthorized party and certain systems were encrypted.
While the Company has restored its core enterprise systems, and critical processes for shipping, receiving, and manufacturing have restarted at some sites with restoration of the remaining sites in process, the timeline for a complete restoration has not yet been finalized.
The 8-K says that data was exfiltrated and some systems encrypted, but not how the data left or how the attacker got in, and that the investigation into the scope and the data affected was ongoing (SEC 8-K, 2026-05-11). No ransomware group had taken credit as of 2026-05-13, and West specialises in injectable drug packaging, syringe and vial components, containment systems and drug delivery devices (BleepingComputer, 2026-05-13); no cited source describes which European supply lines depend on it or what detection preceded the encryption.
incident12 May 05:00Zsingle-source · victim disclosureOpen finding →
CVE-2026-0300 is a buffer overflow in the User-ID Authentication Portal (Captive Portal) service of PAN-OS that lets an unauthenticated attacker execute code with root privileges on PA-Series and VM-Series firewalls by sending crafted packets; Prisma Access, Cloud NGFW and Panorama are not affected (Palo Alto Networks PSIRT). Palo Alto scores it CVSS 4.0 9.3 where the portal is reachable from untrusted networks and 8.7 where it is restricted to trusted internal addresses, and reports limited exploitation against portals exposed to untrusted addresses or the internet (Palo Alto Networks PSIRT). CISA added it to its Known Exploited Vulnerabilities catalog on 2026-05-06 (CISA KEV catalog, 2026-05-06). Fixed builds are listed for every supported branch: 12.1.4-h5 and 12.1.7; 11.2.4-h17, 11.2.7-h13, 11.2.10-h6 and 11.2.12; 11.1.4-h33, 11.1.6-h32, 11.1.7-h6, 11.1.10-h25, 11.1.13-h5 and 11.1.15; 10.2.7-h34, 10.2.10-h36, 10.2.13-h21, 10.2.16-h7 and 10.2.18-h6; older unsupported versions must move to a supported fixed version, and the advisory's table still marks 12.1.7 with an ETA of 05/28 (Palo Alto Networks PSIRT).
Unit 42 attributes the exploitation to CL-STA-1132, a cluster of likely state-sponsored activity: unsuccessful exploitation attempts began on 2026-04-09, and a week later the attackers achieved code execution and injected shellcode (Unit 42, 2026-05-07). They then cleared crash kernel messages and nginx crash entries and records, deployed the tunnelling tools EarthWorm and ReverseSocks5, enumerated Active Directory with the firewall's service-account credentials, and deleted ptrace injection evidence; on 2026-04-29 a SAML flood against the compromised device promoted a second device to active, inheriting the same internet-facing traffic, and that device was then targeted too (Unit 42, 2026-05-07).
Limited exploitation has been observed targeting Palo Alto Networks User-ID™ Authentication Portals that are exposed to untrusted IP addresses and/or the public internet.
Palo Alto Networks released the first wave of fixed PAN-OS builds on 2026-05-13, covering the 12.1, 11.2, 11.1 and 10.2 branches; the advisory still lists 12.1.7 with an ETA of 2026-05-28 (Palo Alto Networks PSIRT). Prisma Access, Cloud NGFW and Panorama remain unaffected, and Threat ID 510019 remains the interim control for unpatched devices on PAN-OS 11.1 or later (Palo Alto Networks PSIRT). CISA lists the flaw in its Known Exploited Vulnerabilities catalog (CISA KEV catalog, 2026-05-06).
Fixed builds are now listed for every supported PAN-OS branch, so the interim mitigation is no longer the only option (Palo Alto Networks PSIRT). Unit 42's date of 2026-04-09 marks the first unsuccessful exploitation attempts; the attackers achieved code execution a week later, and Unit 42 attributes the activity to a single likely state-sponsored cluster, CL-STA-1132, with exploitation described as limited (Unit 42, 2026-05-07). Threat ID 510019 requires PAN-OS 11.1 or later (Palo Alto Networks PSIRT).
Instructure paid ShinyHunters; double Canvas intrusion confirmed; per-institution leak deadline is today (2026-05-12). Ransom acknowledged and "shred logs" received for the platform-wide dataset; a second intrusion on 2026-05-07 defaced ~330 institution portals via the same Free-for-Teacher flaw, and ShinyHunters has now set a fresh per-institution payment deadline (The Register, 2026-05-12). European universities reliant on Canvas should treat the platform-wide settlement as legally unverifiable destruction.
On 2026-05-11 Instructure said it had reached a deal with the actor and received "digital confirmation of data destruction (shred logs)", without disclosing the deal's monetary value (Inside Higher Ed, 2026-05-11). Instructure also confirmed the second intrusion of 2026-05-07, in which about 330 school login portals were defaced through the same Free-for-Teacher vulnerability (The Register, 2026-05-12). ShinyHunters had set a final deadline of end of day 2026-05-12 for individual institutions to contact it, and Instructure's statement says no customers will be extorted as a result of the incident (The Register, 2026-05-12); (Inside Higher Ed, 2026-05-11). CrowdStrike is assisting with the forensic analysis (The Register, 2026-05-12).
Google Threat Intelligence Group's AI Threat Tracker of 2026-05-11, drawn from Mandiant incident response engagements, Gemini and GTIG's own research, follows its February 2026 report on AI-related threat activity (GTIG, 2026-05-11).
An AI-developed zero-day prepared for mass exploitation
GTIG observed prominent cybercrime actors partnering to plan a mass vulnerability-exploitation operation around a zero-day exploit, implemented in a Python script, that bypasses two-factor authentication on a popular open-source, web-based system administration tool; GTIG worked with the vendor to disclose the flaw and disrupt the activity, and says its counter-discovery may have prevented the exploit's use (GTIG, 2026-05-11). The bypass requires valid user credentials and stems from a high-level semantic logic flaw, a trust assumption the developer hardcoded, rather than from memory corruption or input sanitisation (GTIG, 2026-05-11). GTIG does not believe Gemini was used but has high confidence that an AI model supported the discovery and weaponisation, citing abundant educational docstrings, a hallucinated CVSS score and a textbook Pythonic structure with detailed help menus and a clean _C ANSI colour class (GTIG, 2026-05-11). Google did not name the tool (The Hacker News, 2026-05-11). Contradiction: The Hacker News calls this the first time AI was put to use in the wild for exploit generation (The Hacker News, 2026-05-11), whereas GTIG says the planned mass exploitation may have been prevented (GTIG, 2026-05-11).
State actors using AI for vulnerability research and infrastructure
UNC2814 prompted Gemini with expert security personas to support vulnerability research into embedded targets including TP-Link firmware and Odette File Transfer Protocol (OFTP) implementations, and APT45 sent thousands of repetitive prompts that recursively analyse different CVEs and validate proof-of-concept exploits (GTIG, 2026-05-11). APT27 used Gemini to develop a fleet-management application likely meant for an operational relay box network, with the hop count hardcoded to three and mobile Wi-Fi and router devices among the supported types (GTIG, 2026-05-11). Actors also prompt models for the organisational hierarchies and third-party relationships of large enterprises to tailor phishing lures, one tried to identify the exact computer model a high-value target used, and a suspected PRC-nexus actor ran the agentic tools Hexstrike and Strix against a Japanese technology firm and an East Asian cybersecurity platform (GTIG, 2026-05-11).
Malware with AI-generated decoys and an AI-driven Android backdoor
Russia-nexus activity against Ukrainian organisations delivers CANFAIL and LONGSTREAM, which use LLM-generated decoy code to obfuscate their malicious functionality: CANFAIL's source carries LLM comments labelling unused filler blocks, and LONGSTREAM queries the system's daylight-saving status 32 times to pad the script with benign-looking activity (GTIG, 2026-05-11). PROMPTFLUX and HONESTCUE, which call the Gemini API to generate or obfuscate code, featured in GTIG's earlier reports (GTIG, 2026-05-11). The Android backdoor PROMPTSPY, first identified by ESET, runs an agent module that serialises the visible user interface through the Accessibility API, sends it to the gemini-2.5-flash-lite model in an HTTP POST and turns the JSON reply into simulated taps and swipes; it can replay captured PINs or lock patterns, lays an invisible overlay over the uninstall button and can rotate its Gemini API keys and VNC relay through its C2 (GTIG, 2026-05-11).
Pooled LLM access and attacks on AI components
UNC6201 tried a public script that registers and immediately cancels premium LLM accounts, and UNC5673, a PRC-nexus cluster with notable overlaps with TEMP.Hex that has targeted government sectors mainly in South and Southeast Asia, uses Claude-Relay-Service and CLI-Proxy-API to pool Gemini, Claude and OpenAI accounts (GTIG, 2026-05-11). TeamPCP (UNC6780) gained initial access through compromised PyPI packages and malicious pull requests to GitHub repositories including those of Trivy, Checkmarx, LiteLLM and BerriAI, embedded the SANDCLOCK credential stealer to extract AWS keys and GitHub tokens from build environments and monetised them with ransomware and data-theft extortion groups; GTIG adds that the LiteLLM compromise could expose considerable AI API secrets (GTIG, 2026-05-11).
For the first time, GTIG has identified a threat actor using a zero-day exploit that we believe was developed with AI. The criminal threat actor planned to use it in a mass exploitation event but our proactive counter discovery may have prevented its use.
The vulnerability can be classified as a 2FA bypass, though it requires valid user credentials in the first place.
Analysis confirms the use of CANFAIL and LONGSTREAM, which utilize LLM-generated decoy code to obfuscate their malicious functionality.
For example, they employ "Claude-Relay-Service" to aggregate multiple Gemini, Claude, and OpenAI accounts, enabling account pooling and cost-sharing.
TeamPCP gained initial access through compromised PyPI packages and malicious pull requests to these GitHub repositories.
GTIG reports the first zero-day exploit it believes was developed with AI, prepared by criminals for a planned mass-exploitation event that its counter-discovery may have prevented, not an exploit confirmed in use in the wild; the 2FA bypass also requires valid user credentials (GTIG, 2026-05-11). CANFAIL and LONGSTREAM do not call an LLM at runtime: they carry LLM-generated decoy code to obfuscate their functionality, and the report's example of malware driven by an LLM at runtime is the Android backdoor PROMPTSPY (GTIG, 2026-05-11). UNC5673 overlaps with TEMP.Hex rather than being it, and uses relay tooling to pool LLM accounts; the report does not describe stolen LLM API keys as a primary objective or a criminal market for them, and its only point on AI API secrets is that the LiteLLM compromise could expose them (GTIG, 2026-05-11). TeamPCP entered through compromised PyPI packages and malicious pull requests to GitHub repositories, and embedded SANDCLOCK to take AWS keys and GitHub tokens from build environments (GTIG, 2026-05-11).
Upgrade every PA-Series and VM-Series firewall with the User-ID Authentication Portal enabled to the fixed build for its branch (for example 12.1.4-h5, 11.2.10-h6, 11.1.13-h5 or 10.2.18-h6, per the advisory table); until then restrict the portal to trusted zones and disable Response Pages on untrusted L3 interfaces, and check any device that was exposed before patching for the Unit 42 post-exploitation behaviours.
2026-05-12-cd1ab844· Claude Opus 4.7 · 8 entries published
Items dropped or held back
Ivanti EPMM May 2026 chain (CVE-2026-6973 et al.), § 4 UPDATE candidate, dropped after verifier check. S2 surfaced CERTFR-2026-ACT-021, 2026-05-11 framed around named Dutch public-sector victims (Autoriteit Persoonsgegevens, Raad voor de Rechtspraak) as the in-window delta justifying re-coverage. The Phase 5.7 verifier's re-fetch of CERT-FR ACT-021 (and NCSC-NL, Casus kwetsbaarheden Ivanti EPMM-systemen) confirms neither source names those organisations; CERT-FR's affected-systems table reads "Non spécifié par l'éditeur". Without the named-victims claim, the bulletin's only operational delta is a recommendation to rotate EPMM admin credentials as carry-over from the January 2026 chain, already covered in the 2026-05-09 daily. Per PD-8 + PD-13, no fresh material delta in window; dropped. The hallucinated named-victims claim is an example of research-sub-agent drift that the cold-reader verification caught at iteration 1.
CVE-2024-1708 / CVE-2024-1709, ConnectWise ScreenConnect path-traversal + auth-bypass chain (KEV deadline 2026-05-12). S1 surfaced again with Kimsuky (DPRK) / Storm-1175 (China-linked) / Medusa attribution. Both the underlying KEV addition (2026-04-28) and the actor-attribution article (The Hacker News, 2026-04-28) sit outside the 36 h recency window and were already considered and dropped in yesterday's brief on the same basis. Per PD-13, a KEV deadline arriving today is not in itself material new development. No fresh in-window ITW evidence, no fresh in-window research delta. Dropped.
CVE-2026-0073, Android adbd wireless-ADB authentication-bypass (PoC public ~2026-05-11; vendor patch 2026-05-01). S1 surfaced with Penligent/Barghest research (in-window PoC disclosure) and the Android Security Bulletin May 2026. Strict § 2 gate not cleared: vector is AV:A adjacent-network (over wireless ADB on TCP/5555), not internet-exposed; CVSS 8.8 below the ENISA EUVD CVSS-9 threshold; no ITW; not on the KEV catalogue at time of run. Dropped from § 2; logged here for the next brief if exploitation evidence emerges. The mitigation guidance is straightforward and not novel; MDM policies should disable wireless ADB globally on enrolled devices regardless of this CVE.
GLPI cluster (CVE-2026-32312 / 40108 / 42317 / 42318 / 42320 / 42321 / 5385). Already fully treated as § 2 entries in the 2026-05-10 daily; the CERT-FR ACT-021 weekly recap on 2026-05-11 does not add a material delta, instance counts, sectoral framing, and patch paths are unchanged. No re-coverage.
SEPPmail Secure Email Gateway cluster (CVE-2026-44128 / 44125 / 44126 / 44127 / 44129 / 7864). Already the 2026-05-09 deep dive; S2 acknowledged the NCSC-CH security-hub post 12551 is ~44 h old and outside the strict 36 h window. No re-coverage.
CB Financial Services (CBFV) SEC Form 8-K Item 1.05 (filing 2026-05-11 per S4 SEC EDGAR search). S4 could not retrieve the filing body (URL guess fell to 404); no corroborating press coverage was located. Community bank in Carmichaels, PA, limited systemic EU / Swiss relevance, and the filing-detail gap means we cannot quote the disclosure. Held back for tomorrow's brief once the filing body is reachable.
Akamai SIRT "Shadow AI" LLM-malware post. Surfaced by S3 via search but the article's publication date is undetermined; www.akamai.com is not on the bridge allow-list and WebFetch returned 403. Cannot establish in-window status; dropped.
Single-source / reduced-confidence items
West Pharmaceutical Services SEC 8-K is [SINGLE-SOURCE-OTHER] for press corroboration: the SEC filing itself is HIGH-reliability primary, but no tier-1 security press coverage was retrieved before the brief was composed (S4 found only FrenchBreaches.com, an aggregator). Confidence on the disclosure itself: HIGH. Confidence on attribution class (ransomware vs. other actor): MEDIUM; the SEC filing's wording strongly implies double-extortion ransomware but the company does not use the word. Included on the strength of the primary regulator filing.
Škoda Auto online shop breach is corroborated by SecurityWeek + Škoda's own vendor statement; no independent security-research source (Heise, Inside-IT, Le Monde, BKA / BfDI / Czech UOOU enforcement notice) was found before composition. Confidence on the disclosure itself: HIGH. Confidence on the data-exfiltration scope: MEDIUM by Škoda's own admission (logging gap).
ConnectWise ScreenConnect note (above) also relies on cybernews.com and the vendor PSIRT URL, both of which failed to fetch in this run; we are not including the item, but we note that the trust-cert issue on the ConnectWise PSIRT URL is a transport-side recurring problem that we have not yet logged for the bridge fetcher allow-list expansion.
Contradictions / ambiguities
Instructure Canvas "data destruction" claim from ShinyHunters. Instructure's own communication frames receipt of "shred logs" as digital confirmation of deletion; security-research community framing in The Register and Infosecurity Magazine is that ransomware-actor "shred log" artefacts are unverifiable. We follow the security-research framing in the brief because (a) it matches established practice for breach-notification scope decisions under GDPR / Swiss DSG, and (b) the second intrusion on 2026-05-07 and the per-institution deadline reset for today demonstrate that the actor's word is not a forensic primitive.
S2 (Claude Sonnet 4.6, 572 s): returned 3 items. 1 item kept as § 4 UPDATE (Ivanti EPMM CERT-FR ACT-021 + named EU public-sector victims); 2 items already-covered (GLPI 2026-05-10; SEPPmail 2026-05-09 deep dive). Bridge-mandatory hosts NCSC-CH security-hub (12548 + 12551) and ENISA EUVD fetched cleanly via the bridge subcommands. Telemetry: webfetch=14, websearch=18, bridge=8.
S3 (Claude Sonnet 4.6, 642 s): returned 5 items. 2 items kept (GTIG AI Threat Tracker as § 5 Deep Dive per PD-9 annual-report rule; TeamPCP Checkmarx Jenkins as § 4 UPDATE). 2 items duplicated with S4 (Canvas/Instructure ransom-paid; Škoda online-shop breach, merged with S4 framing). 1 item duplicated with S1 (Trellix source-code coverage, already covered 2026-05-10). Telemetry: webfetch=18, websearch=18, bridge=4.
S4 (Claude Sonnet 4.6, 556 s): returned 4 items. 3 items kept (ICO South Staffordshire fine; West Pharma SEC 8-K; Škoda online shop). 1 item duplicated with S3 (Canvas/Instructure UPDATE, merged). Telemetry: webfetch=22, websearch=21, bridge=3.
Tool / source issues observed
hub.ivanti.com, downloads.seppmail.com, ccb.belgium.be are not on the tools/fetch_source.py bridge allow-list, bridge attempts on these hosts return refused: host '...' is not in the allow-list. Direct WebFetch returns 403 on hub.ivanti.com. Coverage is recovered via NCSC-CH security-hub, CERT-FR, and CIRCL vulnerability.circl.lu, but adding these three hosts to the bridge allow-list would remove a recurring single-point coverage risk on Ivanti, SEPPmail, and Belgian-CCB advisories.
bleepingcomputer article-level URLs continue to 403 on direct WebFetch; bridge url returns 200 with SPA-empty body. Listing-level URLs work. No change since 2026-05-11.
advisories-ncsc-nl CSAF bridge: S2 recorded 404 on several recent advisory IDs (NCSC-2026-0384, 0399, 0400, 0401); NCSC-2026-0135 (Ivanti EPMM) fetched cleanly. The NameError that broke yesterday's bridge call is gone; the new mode is upstream 404 on advisory IDs that are presumably embargoed or never published. Worth tracking.
research.nccgroup.com is not on the bridge allow-list (S3 attempted via bridge url; refused).
darkreading article-level URLs continue to 403 on direct WebFetch (S3, S4 both hit this).
databreaches-net Cloudflare Managed Challenge confirmed again; WebSearch fallback used per allow-list, no in-window items surfaced.
Coverage gaps: cisa-kev (bridge OK; no new KEV additions in window 2026-05-11 / 2026-05-12); enisa-euvd (bridge OK; recent lastvulnerabilities returned only low-severity XSS, criticals and exploited only items already covered); ncsc-ch-security-hub (bridge OK; latest posts 12548 + 12551 are 2026-05-08 and already absorbed into S2 returns; no new posts in window); bsi-de (RSS OK; in-window items routine, Linux kernel, libxml2, NGINX, IBM); cert-pl, cert-at, csirt-acn-it (not fetched in this run, rotation candidates for tomorrow); cert-eu (no advisory 2026-007 yet, 2026-006 was the last published); ccn-cert-es (demoted source, not fetched); cnil-fr, edpb, aepd, garante (no in-window enforcement / breach decisions found); ico-uk (sitemap bridge fetched; the South Staffordshire action was the only new enforcement in window); inside-it-ch (Cloudflare-blocked, WebSearch fallback used, no S2 items surfaced); databreaches-net (Cloudflare-blocked, WebSearch fallback, no items); sec-edgar (CB Financial CBFV 8-K Item 1.05 filed 2026-05-11 but filing body not retrievable from EDGAR URL guess, see drops above); hub.ivanti.com, downloads.seppmail.com, ccb.belgium.be, research.nccgroup.com, akamai-sirt, bridge allow-list expansion candidates.