CTIPilot
AI-generated · no human review · verify critical claims against the linked source. how it works →
‹Mon · 18 May 2026›
All daily briefs →
Daily brief · UTC day

Monday, 18 May 2026

3 verified findings from 1 run · the settled record for this UTC day, in the classic brief order.

Criticality
Kind
Topic
Region
TL;DR · the day in one read
  1. 01Tycoon 2FA adds device-code phishing: victims approve token issuance to an attacker device on genuine Microsoft pages. eSentire reported on 2026-05-12 a late-April 2026 campaign in which Tycoon 2FA operators, still active after the March 2026 takedown, added an OAuth device-code variant alongside their credential-relay kit: an invoice lure with a Trustifi click-tracking link leads through Cloudflare Workers and encrypted JavaScript layers to a fake Microsoft CAPTCHA and a voicemail page showing a device code the operator's backend obtained. The victim pastes it at microsoft.com/devicelogin, signs in and approves MFA on genuine Microsoft infrastructure, and Microsoft issues Microsoft Authentication Broker tokens to the operator's polling client; eSentire's hunts look for device-code sign-ins and Node.js user agents in Entra ID logs. →
  2. 02Exchange OWA flaw used in Russia-aligned espionage is fixed since June 2026, but its mailbox foothold needs removing. CVE-2026-42897, an exploited cross-site scripting flaw in Outlook Web Access on Exchange 2016, 2019 and Subscription Edition, is fixed in the June 2026 Security Update, and Microsoft recommends the July 2026 update. Proofpoint saw TA488 (Void Blizzard, LAUNDRY BEAR) exploit it from 2026-07-22 with OWAReaper, whose Owner grant to the Default alias on every mail folder survives password resets and re-imaging and needs deliberate removal from the server. →
  3. 03Switzerland-based THORChain lost about $10.7M when a newly joined validator broke its GG20 threshold signing. THORChain, a Switzerland-based cross-chain liquidity protocol, lost about $10.7M on 2026-05-15 from one of its five vaults by its own count (The Record reported one of six vaults, TRM Labs more than USD 11M across nine chains). THORChain says a validator churned in two days earlier exploited its GG20 threshold-signature implementation: a malformed multi-prime Paillier key planted at setup and 864 deliberately failed signing rounds leaked other validators' key shares until the attacker could sign alone, a chain of three weaknesses in its fork of Binance's tss-lib that upstream had fixed by 2024. THORChain said initial indications were that only protocol-owned funds were affected, and TRM Labs had not attributed the theft as of 2026-05-15. →

01Active threats, incidents & disclosures1 item

NOTABLEupdatedNATOB2

THORChain vault drain, about $10.7M from one vault: a newly joined validator exploited GG20 threshold signing (Switzerland-based protocol)

On 2026-05-15 an attacker drained about $10.7M from one of THORChain's five vaults, and the other four were unaffected (THORChain, 2026-05-20); THORChain's post-mortem of 2026-07-03 puts the loss at about $10M (THORChain, 2026-07-03). The Record reported one of six vaults and about $10.7M (The Record, 2026-05-15), and TRM Labs more than USD 11M across nine chains (TRM Labs, 2026-05-15). THORChain, a cross-chain liquidity protocol founded in 2018 and based in Switzerland, said initial indications were that user funds were safe and only protocol-owned funds were affected (The Record, 2026-05-15).

THORChain confirmed on the evening of 2026-05-15 that the attack exploited its GG20 threshold-signature (TSS) implementation to leak key material progressively, and that the attacker was a node operator churned into the validator set two days earlier (THORChain, 2026-05-20). Its post-mortem describes the chain: at vault setup the attacker published a Paillier key built from several small primes instead of two large ones, which passed because the library only screened out a handful of tiny factors; it then deliberately failed the MTA step of signing 864 times over two and a half days, each failure looking like routine node flakiness while leaking a fragment of other validators' key shares, until it could sign outbound transactions alone (THORChain, 2026-07-03). Two further weaknesses made the leak deterministic, a range proof whose two halves were never tied together and a blinding value drawn from far too small a range; all three sat in THORChain's fork (v0.1.6) of Binance's tss-lib, which upstream had fixed as individually minor issues in v2.0.0 (August 2023) and v3.0.0 (2024) (THORChain, 2026-07-03). THORChain says only the Paillier-modulus condition matches public research, the class documented in 2023 as BitForge and TSSHOCK (THORChain, 2026-07-03); the 2023 disclosures include Fireblocks' CVE-2023-33241 on unchecked Paillier moduli (Fireblocks, 2023-08-09) and Verichains' TSSHOCK attacks on threshold-signing proofs and their encoding (Verichains, 2023-08-10).

The solvency checker halted signing and trading on several chains within about 52 minutes, but only after the funds had left, because the attacker signed with the reconstructed key (THORChain, 2026-05-20). Chainalysis traced attacker-linked funds through Monero, Hyperliquid and Arbitrum from late April, including the ETH that bonded the malicious validator, CryptoTimes reports (CryptoTimes, 2026-05-17). TRM Labs had not attributed the theft to any actor as of 2026-05-15, while calling THORChain the bridge of choice for laundering North Korea's largest thefts (TRM Labs, 2026-05-15). Every validator has since upgraded to v3.19.1, and THORChain plans to replace GG20 with DKLS or Schnorr-based signing (THORChain, 2026-07-03).

Triage: failed signing rounds are routine, since validators go offline or miss timing windows; the discriminator is the aggregate, one recently admitted participant failing the same signing step far more often than its peers over days (THORChain, 2026-07-03).

THORChain officials said the investigation into the incident is ongoing but explained that one of their six vaults was compromised, leading to a loss of about $10.7 million.

The Record 2026-05-15

At the time of writing, TRM has not attributed the May 15 exploit to any specific actor.

TRM Labs 2026-05-15

the operator (or a compromised machine acting as the operator) exploited a vulnerability in the GG20 Threshold Signature Scheme implementation. Rather than a single dramatic key compromise, the attack appears to have involved the gradual leakage of vault key material during keygen or signing rounds, the kind of malformed-proof exploitation that the TSSHOCK class of CVEs first put on the industry's radar a few years ago.

CryptoTimes 2026-05-17

On May 15, 2026, THORChain was the victim of a targeted exploit that drained approximately $10.7M from a single vault (out of five).

The attacker was a newly churned node operator who entered the network two days prior and exploited a vulnerability in the GG20 Threshold Signature Scheme.

After 864 repetitions over two and a half days, the attacker had accumulated enough material to sign transactions unilaterally, bypassing TSS.

The only thing that reveals the pattern is the aggregate: a single validator responsible for a wildly disproportionate share of signing failures over a short period of time.

THORChain
Correctionrun 2026-09-30T0639Z-auditsourcesbodyclassificationtechniquessummaryevidenceevent_dateprioritytitleheadline

The analysis above described CVE-2023-33241 as part of the TSSHOCK class. CVE-2023-33241 is Fireblocks' GG18/GG20 Paillier-key disclosure (Fireblocks, 2023-08-09), and TSSHOCK is Verichains' separate set of key-extraction attacks (Verichains, 2023-08-10). Fireblocks' flaw lies in an unchecked Paillier modulus, which it recommends detecting with a suitable zero-knowledge proof, while TSSHOCK exploits weak or insecurely implemented zero-knowledge proofs. Both let a single malicious participant extract key material (Fireblocks, 2023-08-09; Verichains, 2023-08-10).

The account of the attack above was also narrowed to its sources. The Record reports a compromised vault. The malicious validator node was the vector THORChain's first incident update confirmed, according to CryptoTimes, and the GG20 flaw was then the working theory CryptoTimes reported, supported by PeckShield, Cyvers and security teams working with THORChain rather than Chainalysis (CryptoTimes, 2026-05-17). TRM Labs calls THORChain the bridge of choice for laundering North Korea's largest thefts without naming the Lazarus Group or saying such activity dominates (TRM Labs, 2026-05-15).

Correctionrun 2026-10-20T0835Z-auditbodyevidenceheadlinesourcessourcing_notesummarytechniquestitle

The loss came from one of THORChain's five vaults, not six, and THORChain puts it at about $10.7M (THORChain, 2026-05-20) and in its post-mortem at about $10M (THORChain, 2026-07-03); six vaults is The Record's figure (The Record, 2026-05-15) and more than USD 11M is TRM Labs' (TRM Labs, 2026-05-15), so the six vaults and about $11M stated earlier came from those outlets, not from THORChain. The GG20 flaw was not a suspicion: THORChain confirmed on 2026-05-15 that the attack exploited its GG20 TSS implementation to leak key material progressively (THORChain, 2026-05-20), and its post-mortem of 2026-07-03 sets out the exploit chain, a malformed multi-prime Paillier key and 864 deliberately failed signing rounds combined with two further weaknesses in its tss-lib fork (THORChain, 2026-07-03).

threat18 May 05:00Zmulti-sourceOpen finding →
HIGHCVE-2026-42897exploitedupdatedNATOA1

CVE-2026-42897: Exchange OWA cross-site scripting exploited by TA488 with the OWAReaper implant; fixed since the June 2026 Security Update, which does not remove the mitigation

CVE-2026-42897 is a cross-site scripting flaw in Outlook Web Access on Exchange Server 2016, 2019 and Subscription Edition; Microsoft scores it 8.1, marks it exploited (Microsoft Security Response Center, 2026-07-14), and CISA added it to its Known Exploited Vulnerabilities catalog on 2026-05-15 (CISA KEV catalog, 2026-05-15). Exchange Online is not affected (Microsoft Exchange Team Blog, 2026-05-14). Microsoft first protected on-premises servers with mitigation M2, a URL Rewrite configuration delivered by the Exchange Emergency Mitigation Service (Microsoft Learn). The EM service runs on Mailbox servers (installed with the September 2021 CU or later) and needs outbound HTTPS to officeclient.microsoft.com, which must also be excluded from SSL inspection; servers that cannot reach it never receive the mitigation, and Get-ExchangeServer | Format-List Name,MitigationsApplied shows what each server has applied (Microsoft Learn). A server running an Exchange Server build older than March 2023 cannot check for new mitigations at all (Microsoft Exchange Team Blog, 2026-05-14).

The fix shipped in the June 2026 Security Update (Microsoft Exchange Team Blog, 2026-06-09), and Microsoft recommends the July 2026 Security Update, which also addresses the related CVE-2026-55008; installing either does not remove the earlier mitigation, which stays in place until administrators remove it by hand (Microsoft Exchange Team Blog, 2026-06-09) (Microsoft Exchange Team Blog, 2026-07-14) (Microsoft Security Response Center, 2026-07-14). From 2026-07-22 Proofpoint observed TA488 (Void Blizzard, LAUNDRY BEAR) exploiting the flaw against government, telecommunications, finance, hospitality and aerospace targets in the US and Europe with the OWAReaper implant, whose server-side persistence survives password resets and re-imaging (Proofpoint, 2026-07-29); the update below carries the detail.

The Exchange Emergency Mitigation Service will provide mitigation automatically, and is on by default.

Microsoft Security Response Center 2026-07-14

The messages exploit CVE-2026-42897, a vulnerability in Outlook Web Access in which the server does not adequately sanitize HTML in the message body. This allows a loader piece of JavaScript to use the onload= event handler to parse the rest of the message body, assemble a Base64 fragment, and execute it as encoded JavaScript.

This persistent access lives on the server-side and requires deliberate removal from the Exchange server; credential rotation and even full re-imaging of the targeted user's device will not evict the actor.

Proofpoint 2026-07-29

Installing the July 2026 update does not automatically remove already applied CVE-2026-42897 mitigations.

Please note that EM Service will not be able to check for new mitigations if your server is running Exchange Server version older than March 2023

Microsoft Exchange Team Blog 2026-05-14
Updaterun 2026-07-31T0409Z-intelactionsaffected_productscvesentitiesevidencereferencesregionssectorssourcestagstechniquesbody

CVE-2026-42897 was first an Exchange OWA flaw whose interim protection depended on the EM Service auto-mitigation. Two things changed. Proofpoint has now attributed in-the-wild exploitation to a named Russia-aligned actor and published the implant's full mechanics (Proofpoint, 2026-07-29), and the mitigation is no longer the remediation: the security updates are, with the mitigation now something Microsoft says administrators should remove (Microsoft Exchange Team Blog, 2026-07-14). NCSC-CH appended the Proofpoint reporting to its own advisory on 2026-07-30 (NCSC Switzerland, 2026-07-30).

The actor is TA488 (also tracked as Void Blizzard and LAUNDRY BEAR), the Russia-aligned email-espionage actor behind the Zimbra campaign that Proofpoint documented in a joint release with the NSA (Proofpoint, 2026-07-29). Proofpoint assesses OWAReaper as an evolution of that campaign's ZimReaper payload, citing shared code including the same sized, invisible pop-up and highly similar error handling (Proofpoint, 2026-07-29). Campaign activity began 2026-07-22 against government, telecommunications, finance, hospitality and aerospace targets across the US and Europe, using deliberately banal lure subjects with no call to action; Proofpoint says the unusual breadth may have been intentional blending with bulk mail. Its stated infrastructure-creation date of March 2026 precedes Microsoft's May disclosure by two months, which is the basis for its assessment that zero-day use is feasible; that is an inference from infrastructure dating, not a confirmed finding.

Execution. The flaw is a failure to sanitise HTML in the message body, so a loader script in an onload= handler reassembles a Base64 fragment from the rest of the message and evaluates it: no link click and no attachment open, only viewing the message in OWA. The exploit and payload fragments are hidden inside the message's social-media icon elements, with next-stage data placed after # fragment markers where the browser's Base64 image parser stops reading, so the payload is not visible to casual inspection of the HTML. On execution OWAReaper first rewrites the delivered message server-side to strip the exploit content and suppresses OWA pop-ups and right-click, then enumerates the victim's address, username and settings (Proofpoint, 2026-07-29).

Credential and token theft. It creates two invisible input elements and waits for the browser's own autofill to populate them with the saved OWA username and password. Separately it enumerates installed Outlook add-ins holding ReadWriteMailbox permission and, where one exists, abuses it to call GetClientAccessToken and obtain OAuth tokens (Proofpoint, 2026-07-29).

Persistence, in three independent layers. Client-side, the implant writes an encrypted copy of itself and a decryption wrapper into browser localStorage under a settings field of the legitimate PageDataPayload.OwaUserDefaultSettings key, which OWA itself evaluates during its own sync-restore flow, so every ordinary OWA tab-open re-launches it with no separate loader. A second client-side layer adds a hidden iframe to messages cached in OWA's offline IndexedDB store, so opening the cached message re-infects an endpoint even after a full re-image. The third is server-side and is the one that matters most: the implant calls UpdateFolder to grant Owner-level permission on every mail folder to the low-privilege "Default" preset alias that exists in every Exchange organisation. Proofpoint is explicit that this "requires deliberate removal from the Exchange server" and that credential rotation and re-imaging will not evict it (Proofpoint, 2026-07-29).

Command and control. Two channels, both over infrastructure defenders generally trust. The implant polls GitHub's public Commit Search API every 24 hours for crafted commit messages containing the target's own email address, decrypting matches to a four-character command header that selects toolkit replacement, C2-domain rotation, or one-off code execution; in parallel it re-parses cached inbound messages every five minutes for the same command structure. Exfiltration runs primarily over HTTPS with encrypted URI paths, either relayed through a set of legitimate image-CDN domains or sent directly to the actor-controlled server when those proxies fail; if the HTTPS method fails altogether, the implant switches to DNS label tunnelling, packing the data into the subdomain labels of ordinary DNS queries for an actor-controlled domain. Notably, Proofpoint states there is no mass mailbox exfiltration here, unlike the Zimbra campaign, so the mapped behaviour is browser-session and credential access rather than bulk email collection (Proofpoint, 2026-07-29).

Patching. The fix first shipped in the June 2026 Security Update (Microsoft Exchange Team Blog, 2026-06-09), and Microsoft now recommends the July 2026 Security Update, available as Exchange SE RTM publicly and for Exchange 2019 CU14/CU15 and Exchange 2016 CU23 only through the Period 2 Extended Security Update programme; organisations that were enrolled only in Period 1, which ended in April 2026 (Microsoft Exchange Team Blog, 2026-05-14), do not receive it (Microsoft Exchange Team Blog, 2026-07-14). Microsoft's own vulnerability record scores the flaw 8.1 and marks it exploited (Microsoft Security Response Center, 2026-07-14). Installing the update does not remove a previously applied mitigation: administrators who used the EM Service should remove the M2.1.0 IIS rules through the documented rollback, and those who ran the downloadable mitigation script should run its rollback (Microsoft Exchange Team Blog, 2026-07-14). The known operational side effects of the mitigation era (broken OWA calendar printing, inline-image rendering problems, OWA-light failing, and false-unhealthy calendar-proxy health alerts) are listed in the May post (Microsoft Exchange Team Blog, 2026-05-14), and the July post ties their clearing to installing the update and removing the mitigation (Microsoft Exchange Team Blog, 2026-07-14).

Triage: OWA legitimately reads and writes its own settings keys constantly, so the presence of localStorage activity is not the signal; the discriminator is the specific settings-field path carrying an encrypted blob, and its correlation with a message open. For the server-side artifact the discrimination is cleaner: administrators do grant folder permissions, but they grant them to named users or groups for a specific folder, not Owner rights to the built-in "Default" alias across an entire mailbox. Treat any such grant as compromise until proven otherwise.

Improvementrun 2026-08-30T1312Z-auditactionsclassification

The sourcing is the vendor's own advisory for its own product, independently corroborated: Proofpoint analysed the implant separately from Microsoft's disclosure, and NCSC-CH's post added a link to Proofpoint's analysis. Nothing in the assessment or the remediation guidance changed.

Correctionrun 2026-10-11T1307Z-audittitleheadlinesummarypriorityimmediate_actiontagscvessourcesevidencesourcing_noteactionsbodyevent_datesectors

The EM service fetches mitigations from officeclient.microsoft.com over port 443, runs only on Mailbox servers, and reports applied mitigations through Get-ExchangeServer | Format-List Name,MitigationsApplied (Microsoft Learn); the earlier host name and Edge Transport cmdlet had no source. The fix first shipped in the June 2026 Security Update (Microsoft Exchange Team Blog, 2026-06-09), a month before the July update Microsoft now recommends (Microsoft Exchange Team Blog, 2026-07-14). Proofpoint describes its July 2026 report on the actor's Zimbra campaign as a joint release with the NSA (Proofpoint, 2026-07-29).

Improvementrun 2026-10-21T1149Z-auditbodyevidencetechniquesheadline

The Exchange Emergency Mitigation Service cannot check for new mitigations on a server running an Exchange Server build older than March 2023 (Microsoft Exchange Team Blog, 2026-05-14), so such a server did not receive mitigation M2 automatically.

Builds on: 16-nation advisory: Russia's LAUNDRY BEAR exfiltrates government mail through a view-based…

vulnerability18 May 05:00Zmulti-sourceOpen finding →

03Deep dive1 item

HIGHupdatedNATOB2

Tycoon 2FA after the March 2026 takedown: operators add OAuth device-code phishing against Microsoft 365 alongside the credential-relay kit

eSentire's Threat Response Unit analyzed in late April 2026 a campaign by operators of the Tycoon 2FA phishing-as-a-service kit, which kept running after the March 2026 takedown led by Microsoft and Europol, that keeps the kit's in-browser tradecraft but ends in Microsoft's legitimate device-login flow instead of a credential-relay proxy (eSentire TRU, 2026-05-12). The device-code variant runs beside the relay kit: eSentire still names the OfficeHome AppId as the primary marker of the credential-relay variant and treats the two AppIds as one detection set (eSentire TRU, 2026-05-12), and BleepingComputer reports that the kit now supports device-code phishing (BleepingComputer, 2026-05-17).

The lure, a forwarded vendor-invoice reminder, carries a Trustifi click-tracking URL whose clean-reputation domain launders the link, and Trustifi redirects to a throwaway Cloudflare Workers subdomain that returns an AES-GCM-encrypted page decrypted in the browser (eSentire TRU, 2026-05-12). The decrypted controller blanks the page for headless browsers and Burp, suppresses developer-tool shortcuts, redirects to microsoft.com when a debugger timing trap fires, checks the visitor's network owner against 230 cloud, sandbox, security-vendor, VPN and AI-crawler names, and stops rendering after a hardcoded expiry date (eSentire TRU, 2026-05-12). A Microsoft-branded fake CAPTCHA then queries a Tycoon 2FA Check Domain, which serves suspected analysts a benign decoy website and real victims a Microsoft 365 voicemail lure (eSentire TRU, 2026-05-12).

The voicemail page displays a user code that the operator's backend obtained from Microsoft's device-code endpoint, copies it and opens microsoft.com/devicelogin; the victim signs in at login.microsoftonline.com, approves the MFA push and consents to Microsoft Authentication Broker, and Microsoft issues access and refresh tokens to the operator's polling client, after which the kit sends the victim to their real Outlook inbox (eSentire TRU, 2026-05-12). A victim already signed in sees no authentication step at all, only the consent prompt (eSentire TRU, 2026-05-12). Because the Broker is a Microsoft first-party application it does not trigger unverified-publisher OAuth detections, and its token can be exchanged for Exchange, Graph and OneDrive tokens without further interaction (eSentire TRU, 2026-05-12). The backend traffic still uses the CryptoJS AES-CBC layer with a hardcoded key and IV unchanged since eSentire's April 2025 analysis, which works as a kit fingerprint rather than protection, and the operators' token traffic came from Alibaba (US) hosting, which eSentire has seen across both kit variants since about 2026-04-10 in addition to the hosting providers documented before (eSentire TRU, 2026-05-12).

Triage: a device-code sign-in alone can be a legitimate command-line tool or input-constrained device; eSentire's discriminators are a source with no prior device-code activity and Node.js user agents against the Broker AppId in non-interactive sign-ins, which it calls highly anomalous in production environments (eSentire TRU, 2026-05-12).

Doing so authorizes the attacker to register a rogue device with the victim’s Microsoft 365 account, giving them unrestricted access to the victim's data and services, including email, calendar, and cloud file storage.

BleepingComputer 2026-05-17

The user's MFA worked exactly as designed. There is no proxy, no credential capture, no fake Microsoft page; everything from login.microsoftonline.com onward is authentic Microsoft infrastructure responding to authentic Microsoft authentication events.

Successful device-code authentication (AuthenticationProtocol = deviceCode, ResultType = 0) from sources with no prior device-code activity.

eSentire Threat Response Unit 2026-05-12

We recommend organizations get as close as possible to a unilateral block on device code flow.

Microsoft Learn
Correctionrun 2026-10-21T1148Z-auditaffected_productsbodyclassificationentitiesevent_dateevidenceheadlineregionssectorssourcessourcing_notesummarytagstechniquestitleverification

Tycoon 2FA operators did not move away from the credential-relay kit: eSentire still names the OfficeHome AppId as the primary marker of the relay variant and treats it and the Broker AppId as one detection set, and BleepingComputer reports that the kit now supports device-code phishing (eSentire TRU, 2026-05-12; BleepingComputer, 2026-05-17). The Alibaba (US) hosting seen since about 2026-04-10 is an addition to the hosting providers documented before, not a replacement (eSentire TRU, 2026-05-12). The lure page shows a user code obtained by the operator's backend and tells the victim to paste it at microsoft.com/devicelogin (BleepingComputer, 2026-05-17), after which the victim sees a consent prompt reading "Are you trying to sign in to Microsoft Authentication Broker?" (eSentire TRU, 2026-05-12). The detection signals eSentire publishes are successful deviceCode sign-ins from sources with no prior device-code activity, tokens issued to the Microsoft Authentication Broker AppId, node or undici user agents in non-interactive sign-ins, and ResultType 50173 after revocation (eSentire TRU, 2026-05-12). eSentire says the phish does not bypass MFA but changes what MFA authorizes, and recommends blocking the device-code flow with Conditional Access as the control that removes the attack class (eSentire TRU, 2026-05-12), in line with Microsoft's advice to get as close as possible to a unilateral block (Microsoft Learn); no source says that FIDO2 sign-in defeats device-code phishing.

threat18 May 05:00Zsingle-sourceOpen finding →

04Action items2 items

Verification & coverage notes1 run

2026-05-18-2eabc1cf · Claude Opus 4.7 · 5 entries published

  • Coverage window: standard daily (gap to prior brief 2026-05-17 ≈ 24 h; window_hours = 36). Quiet Sunday-into-Monday, § 2 and § 3 are intentionally empty per PD-11.
  • Items dropped (sub-agent returned but failed Phase 2 / dedup / recency):
    • SEPPmail CVE-2026-44125 / 44126 / 44127 / 44128 / 44129 / 7864 cluster (NCSC-CH post #12551, 2026-05-08); already covered in the 2026-05-09 deep dive and the CVEs are all in cves_seen.json; the NCSC-CH advisory date (2026-05-08) is 10 days outside window_hours = 36; dropped, no in-window delta.
    • Windows YellowKey / GreenPlasma zero-days (NCSC-CH post #12574, 2026-05-14), already covered in the 2026-05-15 § 1; no fresh in-window development.
    • DHTMLX CVE-2026-41553 / 41552 / 7182, already covered as a TL;DR item in 2026-05-17; CERT-PL advisory date 2026-05-15 sits at the edge of window but the coverage is already current.
    • NCSC-CH weekly review Week 19 (advance-fee scam, double-phishing awareness items); primary-source date 2026-05-12 is outside window_hours; awareness-class content with no fresh defender action.
  • Single-source items: [SINGLE-SOURCE] CVE-2026-0300 PAN-OS § 4 UPDATE; sole primary source is the Palo Alto Networks PSIRT advisory (vendor-authoritative; national-CERT carve-out does not apply but vendor-PSIRT is itself the primary disclosing party).
  • Included with reduced confidence (no in-window primary): none in this brief; the three out-of-window S3 items were dropped rather than promoted.
  • Out-of-window research deferred to weekly summary (or next-week daily if material develops):
    • The DFIR Report, 2026-05-11, EtherRAT blockchain-C2 + TukTuk SaaS-C2 chain ending in Gentleman ransomware; novel detection-engineering content (EtherHiding / Arweave dead-drop / multi-SaaS C2 fingerprints) but source is 7 days outside window_hours.
    • Microsoft Security Blog, 2026-05-12, 123-day MSP-mediated intrusion via HPE Operations Manager with malicious Windows Network Provider DLL and LSA password-filter persistence; high relevance to public-sector outsourced IT but 6 days outside window_hours.
    • Unit 42, 2026-05-11, Active Directory Certificate Services ESC1 + shadow-credential exploitation attributed to Fighting Ursa (APT28); 7 days outside window_hours.
  • Contradictions surfaced: CVSS scoring for CVE-2026-42945 NGINX Rift differs across primaries; NCSC-CH lists CVSS 4.0: 9.2 Critical while NVD currently has no published score; The Hacker News and Security Affairs cite the F5 advisory's CVSS 4.0 base of 9.2 used in this brief. CVSS 3.1 score reported by NCSC-NL feed is 8.1. Brief uses the CVSS 4.0 score most widely cited by national-CERT sources.
  • Sub-agents that didn't return on time: none, S1 (348 s), S2 (677 s), S3 (576 s), S4 (679 s) all returned inside the 30-min hard cap.
  • Candidate sources surfaced this run (one new candidate maximum per PD-3.6): depthfirst (depthfirst.com), AI-assisted vulnerability research, primary disclosure source for CVE-2026-42945 NGINX Rift cited by NCSC-CH; recorded as status: candidate in sources/sources.json. A second candidate (cryptotimes) was surfaced by S4 for THORChain technical post-mortems and is held for a future run per the one-candidate-per-run cap.
  • Coverage gaps: cisa-kev (bridge subcommand returned no in-window adds beyond CVE-2026-20182 / CVE-2026-42897 already covered); apple-security (no in-window emergency update); chrome-releases (no in-window emergency update); akamai-sirt (RSS 403, no in-window content corroborated via search); trendmicro-research (feed parse error, no in-window content corroborated via search); sophos-xops (HTTP 503, no in-window content corroborated via search); inside-it-ch (host 403, no in-window CH-specific incidents); cert-eu (most recent advisory 2026-05-06; feed empty in window); ncsc-ch weekly-review-kw20 (Week 20 review not yet published as of 2026-05-18T04:50Z); cert-fr actu (feed appears stale, items dated Sep–Oct 2025); sec-disclosures-edgar (Item 1.05 search returned zero filings for 2026-05-15 → 2026-05-18, US weekend); ico-uk (no new enforcement actions in window); cnil-fr (no new enforcement decisions in window); databreaches-net (host 403, WebSearch fallback used per documented mitigation).

Unmatched action items (migrated)

  • Block OAuth Device Code flow tenant-wide in Entra ID Conditional Access where it is not operationally required. Path: Conditional Access → New policy → Conditions → Authentication flows → Device code flow → Block. Scope to all user accounts and exempt only the named service principals (smart-TV, IoT, CLI) that demonstrably need it. Where a wholesale block is infeasible, restrict the device-code flow to compliant devices and named-location IPs. Monitor Entra ID sign-in logs for AuthenticationProtocol = "deviceCode" from unfamiliar IPs against high-privilege users, see § 5 deep dive.

Migrated from briefs/2026-05-18.md (v2).