Last Week in Threats Banner

Last Week in Threats: Week 38

Bottom Line

What changed this week: Two actively exploited Cisco vulnerabilities landed in the KEV catalog, both targeting enterprise perimeter security: a SQL injection flaw in Secure Email Gateway enabling root command execution, and an authentication bypass in Identity Services Engine.

Who is most exposed: Organizations running Cisco email gateways or ISE for identity management, plus technology firms, manufacturers, and healthcare organizations facing a 14% ransomware volume increase driven by Qilin's near-quadruple surge.

Recommended action: Patch CVE-2026-76461 and CVE-2026-76460 immediately, hunt for unauthorized Cisco ISE authentication events and anomalous email gateway command execution, and review outbound data transfers from technology and manufacturing environments.

The Week in Review

The week of September 14 to 20, 2026 delivered a double Cisco vulnerability event targeting enterprise perimeter defenses. Two flaws hit CISA's Known Exploited Vulnerabilities catalog: CVE-2026-76461, a SQL injection in Secure Email Gateway that grants root command execution, and CVE-2026-76460, an authentication bypass in Identity Services Engine. Both are under active exploitation with public proof-of-concept code in circulation, per AlienVault OTX (16 and 13 community pulses respectively, pulled September 21). Organizations relying on these systems for email filtering or network access control faced a compressed mitigation window.

Ransomware activity climbed to 185 leak-site postings (a 14% increase from last week's 162, per ransomware.live, pulled September 21). Qilin surged to the top with 31 victims, nearly quadrupling from 8 the prior week. Technology firms (38 victims), manufacturers (31), and healthcare organizations (22) carried the highest sector concentration. The week's top 15 malicious IPs split evenly between datacenter hosting (7) and compromised residential networks (7), with one Tor exit node, signaling attacks from both infrastructure and home broadband endpoints (AbuseIPDB 100% confidence blacklist, pulled September 21).

Ransomware Leak-Site Activity

The week produced 185 victim postings across ransomware leak sites, a 14% increase from the prior week's 162. Qilin dominated with 31 victims, surging from 8 the week prior (nearly a 4x increase). The Gentlemen held second with 30 postings, up from 18. N0n appeared for the first time with 11 victims. The United States accounted for 58 postings, followed by 23 with unknown geography. Germany, Italy, and Brazil rounded out the top five victim countries. (Source: ransomware.live, aggregated from ledger data via aggregate-week.py, pulled September 21, 2026)

Most active groups by victim postings (ransomware.live, pulled Sep 21, 2026)
qilin 31 thegentlemen 30 N0n 11 akira 10 Storm 8 safepay 8 emperador 7 Panzer 6 lockbit5 5 metaencryptor 5

Sector & Technology Watch

Technology firms bore the highest ransomware exposure this week with 38 victims, followed by manufacturing (31) and healthcare (22). The concentration signals persistent targeting of intellectual property and operational technology environments. The week's exploited vulnerabilities clustered in enterprise perimeter technology: identity and VPN systems (Cisco ISE authentication bypass), email gateways (Cisco Secure Email Gateway SQL injection), backup systems (Acronis cPanel/Plesk privilege escalation), and mobile devices (Google Pixel modem authorization flaw). Organizations running Cisco's email or identity stack faced the most compressed mitigation window. If you operate Cisco ISE or Secure Email Gateway, patch immediately and hunt for authentication anomalies and unexpected command execution in gateway logs.

Most targeted sectors by victim postings (ransomware.live, pulled Sep 21, 2026)
Technology 38 Manufacturing 31 Healthcare 22 Professional Svcs 19 Construction 14 Education 11 Finance 9 Retail 8 Government 7 Transportation 6

Most-Reported Malicious IPs

The top 15 IPs from AbuseIPDB's 100% confidence blacklist (pulled September 21, 2026) split evenly between datacenter hosting (7) and compromised residential broadband (7), with one institutional Tor exit node. The most-reported address, 182.93.7.194 (Macau, 14,817 reports from 1,249 distinct users), originates from a residential ISP (CTM), signaling a compromised home or small-office endpoint rather than malicious infrastructure. Top hosting sources included DigitalOcean (3,560 reports) and Storm Industries CDN (2,433 reports). GreyNoise (community tier, pulled September 21) flagged 6 of the top 10 as malicious scanners and 1 as a benign research scanner (Shadowserver). Observed behavior centered on SSH brute-force (6 IPs), port scanning (4), web application attacks (3), and email/credential brute-force (2).

Note: One IP (65.49.20.82, Shadowserver Foundation) is a benign internet-wide security scanner and should be excluded from blocking rules. The Tor exit node (185.220.101.31) legitimately anonymizes traffic and may warrant a review of your organization's Tor usage policy rather than outright blocking.

Attacker Infrastructure & Networks

The week's top malicious IPs reveal a balanced attack surface: 7 originated from datacenter hosting or content delivery networks, 7 from compromised residential broadband, and 1 from an institutional Tor exit. The residential segment signals widespread home-router and endpoint compromise rather than malicious ISPs. The highest-volume residential source, AS4609 (CTM Macau, 14,817 reports), likely reflects a compromised subscriber device. Among hosting networks, AS14061 (DigitalOcean, 3,560 reports) and AS7029 (Windstream/Storm Industries CDN, 2,433 reports) topped the attacker-network list. No ThreatFox or URLhaus correlation was found for the enriched IPs, meaning none doubled as known malware C2 or payload hosts during the analysis window. (Network ownership: AlienVault OTX; usage type and report volume: AbuseIPDB; malware-infra correlation: ThreatFox and URLhaus, all pulled September 21, 2026)

Top networks by count of the week's malicious IPs (AlienVault OTX + AbuseIPDB, pulled Sep 21, 2026)
AS4609 CTM (residential) 1 AS14061 DigitalOcean (hosting) 1 AS7029 Windstream (hosting) 1 AS6939 Hurricane Electric (residential) 1 AS45753 NetSec (hosting) 1 AS8075 Microsoft (hosting) 1 AS4818 DiGi Telecom (residential) 1 AS45013 Econotel (residential) 1 AS209371 Private Network (hosting) 1 AS43754 Asiatech (residential) 1
Origin composition: hosting vs residential vs other (AbuseIPDB usage type, pulled Sep 21, 2026)
Hosting: 7 Residential: 7 1 Hosting/datacenter: 7 | Residential broadband: 7 | Business/institution: 1 | Total enriched IPs: 15

Exploited Vulnerabilities

CISA added four vulnerabilities to the Known Exploited Vulnerabilities catalog during the week of September 14 to 20, 2026. Two Cisco flaws targeting enterprise perimeter systems led the list: CVE-2026-76461, a SQL injection in Secure Email Gateway's AsyncOS enabling unauthenticated root command execution (due date September 17), and CVE-2026-76460, an authentication bypass in Identity Services Engine allowing unauthenticated access to the management interface (due date September 19). Both vulnerabilities are under active exploitation with public proof-of-concept code, per AlienVault OTX (16 and 13 community threat pulses respectively, pulled September 21). The remaining two KEV additions were CVE-2026-87886 (Acronis Backup cPanel/Plesk privilege escalation, due September 19) and CVE-2026-58704 (Google Pixel cellular modem authorization bypass, due September 19).

Malware & Command-and-Control Activity

A 2-day window of high-confidence indicators from ThreatFox (pulled September 21, 2026) captured 2,027 IOCs, including 440 fresh command-and-control nodes. AsyncRAT dominated with 974 indicators, followed by ClearFake (120), VShell (103), Cobalt Strike (101), and an unattributed loader (92). The AsyncRAT volume signals sustained credential-theft and remote-access campaigns. ClearFake and IClickFix (75 IOCs) represent browser-update social engineering chains delivering malware via fake software prompts. Legacy botnet activity persisted with Mirai (74 IOCs). The Cobalt Strike presence (101 IOCs) reflects ongoing post-exploitation framework abuse in intrusion chains.

The Hunting Takeaway

This week's double Cisco vulnerability event creates a clear behavioral hunting target: unauthorized authentication to Cisco Identity Services Engine and anomalous command execution on Secure Email Gateway appliances. Hunt for ISE authentication events where the user or source IP has no prior login history, especially from external or unexpected internal networks. On email gateway appliances, look for AsyncOS process execution outside normal administrative sessions, particularly shell command invocations or unexpected child processes spawned by the email filtering service. Both patterns signal exploitation of CVE-2026-76460 (ISE auth bypass) and CVE-2026-76461 (email gateway SQL injection to root RCE).

Microsoft KQL (Cisco ISE authentication anomaly hunt):

// Hunt for Cisco ISE authentication from previously unseen source IPs
// Requires syslog ingestion from ISE appliances into a Syslog or CommonSecurityLog table
let lookback = 30d;
let recent = 7d;
let knownSources = Syslog
| where TimeGenerated between (ago(lookback) .. ago(recent))
| where SyslogMessage has "CISE" and SyslogMessage has_any ("Passed-Authentication", "authentication succeeded")
| extend SourceIP = extract(@"(\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3})", 1, SyslogMessage)
| where isnotempty(SourceIP)
| distinct SourceIP;
Syslog
| where TimeGenerated > ago(recent)
| where SyslogMessage has "CISE" and SyslogMessage has_any ("Passed-Authentication", "authentication succeeded")
| extend SourceIP = extract(@"(\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3})", 1, SyslogMessage)
| where isnotempty(SourceIP)
| where SourceIP !in (knownSources)
| project TimeGenerated, Computer, SourceIP, SyslogMessage
| order by TimeGenerated desc

Splunk SPL (Cisco Secure Email Gateway command execution anomaly hunt):

index=email sourcetype=cisco:esa
    (process_name=*sh OR process_name=python* OR process_name=perl OR ParentProcessName=*asyncos*)
    earliest=-7d
| eval CommandLine=coalesce(CommandLine, process)
| where isnotnull(CommandLine) AND (like(CommandLine, "%/bin/%") OR like(CommandLine, "%sh -c%") OR like(CommandLine, "%wget%") OR like(CommandLine, "%curl%"))
| stats count min(_time) as FirstSeen max(_time) as LastSeen values(CommandLine) as Commands by host, user, process_name
| convert ctime(FirstSeen) ctime(LastSeen)
| where count < 5
| table FirstSeen, LastSeen, host, user, process_name, Commands, count
| sort -LastSeen

Note: These queries were generated with AI assistance and are a starting point, not a verdict. A query that returns no results does not mean the threat is absent. Confirm that the referenced indexes, sourcetypes, and field names exist in your environment, that the relevant data is actually being ingested, and that names match your schema (watch for spelling and naming drift). Validate against your own ingest and audit trails before drawing conclusions.

Check Your Environment

Use the queries below to sweep your logs for this week's high-confidence malicious IPs and cross-reference the KEV additions against your vulnerability inventory. These are environment sweeps, not behavioral hunts. No results do not mean no threat.

Microsoft KQL (network connection sweep for this week's top malicious IPs):

// Check for connections to this week's 100% confidence AbuseIPDB malicious IPs
// Excludes the benign Shadowserver scanner
let MaliciousIPs = dynamic([
    "64.225.30.2", "103.117.145.130", "40.74.211.76", "182.93.7.194",
    "49.124.149.207", "79.135.251.82", "85.204.70.102", "171.22.18.65",
    "212.33.198.115", "34.35.76.234", "213.124.148.22", "185.220.101.31",
    "103.191.204.229", "94.154.43.84"
]);
union
    (DeviceNetworkEvents
    | where TimeGenerated > ago(7d)
    | where RemoteIP in (MaliciousIPs) or LocalIP in (MaliciousIPs)),
    (CommonSecurityLog
    | where TimeGenerated > ago(7d)
    | where DestinationIP in (MaliciousIPs) or SourceIP in (MaliciousIPs))
| project TimeGenerated, DeviceName, SourceIP=coalesce(LocalIP, SourceIP), DestinationIP=coalesce(RemoteIP, DestinationIP), RemotePort, Action
| order by TimeGenerated desc

Splunk SPL (firewall/proxy sweep for this week's malicious IPs):

index=network OR index=proxy
    (src_ip IN ("64.225.30.2", "103.117.145.130", "40.74.211.76", "182.93.7.194", "49.124.149.207", "79.135.251.82", "85.204.70.102", "171.22.18.65", "212.33.198.115", "34.35.76.234", "213.124.148.22", "185.220.101.31", "103.191.204.229", "94.154.43.84")
     OR dest_ip IN ("64.225.30.2", "103.117.145.130", "40.74.211.76", "182.93.7.194", "49.124.149.207", "79.135.251.82", "85.204.70.102", "171.22.18.65", "212.33.198.115", "34.35.76.234", "213.124.148.22", "185.220.101.31", "103.191.204.229", "94.154.43.84"))
    earliest=-7d
| table _time, src_ip, dest_ip, dest_port, action, bytes
| sort -_time

Vulnerability check: Cross-reference CVE-2026-76461 (Cisco Secure Email Gateway), CVE-2026-76460 (Cisco ISE), CVE-2026-87886 (Acronis Backup), and CVE-2026-58704 (Google Pixel) against your asset inventory. Patch immediately where present.

Note: These queries were generated with AI assistance and are a starting point, not a verdict. A query that returns no results does not mean the threat is absent. Confirm that the referenced indexes, sourcetypes, and field names exist in your environment, that the relevant data is actually being ingested, and that names match your schema (watch for spelling and naming drift). Validate against your own ingest and audit trails before drawing conclusions.

Appendix: Raw Data Tables

The tables below are the tabular view of the charts above. Each table carries its source and pull date. Click column headers to sort; use the Copy button to export.

Top Ransomware Groups (ransomware.live, pulled Sep 21, 2026)

Group Victims This Week Victims Last Week Change
qilin318+287%
thegentlemen3018+67%
N0n110NEW
akira107+43%
Storm80NEW
safepay811-27%
emperador76+17%
Panzer60NEW
lockbit550NEW
metaencryptor50NEW

Most-Reported Malicious IPs (AbuseIPDB 100% confidence, pulled Sep 21, 2026)

IP Address Country Total Reports Distinct Reporters Origin Type ASN
182.93.7.194MO14,8171,249ResidentialAS4609
64.225.30.2US3,560963HostingAS14061
94.154.43.84NL2,433535HostingAS7029
65.49.20.82US2,020147Residential*AS6939
103.117.145.130HK1,676694HostingAS45753
85.204.70.102FR1,515453HostingN/A
103.191.204.229IN982256ResidentialN/A
79.135.251.82RU648158ResidentialAS45013
171.22.18.65DE578337HostingAS209371
213.124.148.22NL512134ResidentialAS33915
49.124.149.207MY403106ResidentialAS4818
185.220.101.31DE228105Business†AS208294
212.33.198.115IR212172ResidentialAS43754
40.74.211.76US16569HostingAS8075
34.35.76.234ZA7152HostingN/A

*Shadowserver Foundation benign research scanner. †Tor exit node operated by Artikel10 e.V.

New CISA KEV Additions (CISA KEV catalog, pulled Sep 21, 2026)

CVE Vendor/Product Vulnerability Date Added Due Date
CVE-2026-76461Cisco Secure Email GatewaySQL Injection (root RCE)Sep 14, 2026Sep 17, 2026
CVE-2026-76460Cisco Identity Services EngineAuth BypassSep 16, 2026Sep 19, 2026
CVE-2026-87886Acronis BackupPrivilege EscalationSep 16, 2026Sep 19, 2026
CVE-2026-58704Google PixelAuthorization BypassSep 16, 2026Sep 19, 2026

Top Malware Families (ThreatFox 2-day window, pulled Sep 21, 2026)

Malware Family IOC Count
AsyncRAT974
ClearFake120
VShell103
Cobalt Strike101
Unknown Loader92
Unknown malware81
IClickFix75
Mirai74
Vidar59
php.shin_webshell51

Methodology & Sources

This recap aggregates open-source threat intelligence feeds captured during the week of September 14 to 20, 2026, and cross-references them to surface trends and behavioral hunting angles. The figures are aggregate public statistics, not Focused Hunts detections, and represent a snapshot of the threat landscape as visible through these sources at the time of collection. Each section cites its upstream provider and pull date inline; the full data is retained in a timestamped snapshot file for audit and reproducibility.

Week-over-week trend comparisons appear once successive snapshots exist. When ledger coverage for a given week is incomplete, the total is reported as a floor with the covered sub-range disclosed. Unavailable source classes are omitted rather than estimated. Actor claims and confirmed intelligence listings (such as CISA KEV) are kept separate and labeled as such. The hunting queries are behavioral starting points, not verdicts; they require validation against your own telemetry and schema. A query that returns no results does not mean the threat is absent.