ctxsec

SAP October Patch Day: CVSS 10.0 OAuth2 SSRF in Cloud SDK for Python (CVE-2026-89111)

Tue, 13 Oct 2026 00:00:00 +0000

SAP's Oct 13 Security Patch Day ships 16 new notes and 4 GitHub advisories. Highest severity: CVE-2026-89111 (CVSS 10.0) in SAP Cloud SDK for Python <0.45.2 — attacker-controlled tenant_subdomain can redirect OAuth2 token requests and exfiltrate client credentials in multi-tenant BTP apps. Fix: upgrade to 0.45.2+; workaround: allowlist tenant subdomains. Also Critical: NetWeaver Visual Composer broken access control (CVE-2026-89110, 9.9) and S/4HANA Manage Substitution/Validation Rules code injection (CVE-2026-89107, 9.9).

Meta: stolen employee credential used to access GitHub repos; ~1,500 people notified

Tue, 13 Oct 2026 00:00:00 +0000

Meta told SecurityWeek that an outside group used a Meta employee's stolen credentials to access some company GitHub repositories. About 99% of what was taken was business/source code; ~1% was personal data stored against policy (some synthetic/anonymized test data). Roughly 1,500 people — mostly U.S. Meta employees — are being notified. Meta says no Meta product vulnerability and internal systems were not breached; secrets were rotated and new controls added to keep user data out of GitHub. Law enforcement and GitHub were engaged.

XBOW: AI-assisted Chrome full-chain (6 days) earns Google's first $250k 2026 VRP award

Tue, 13 Oct 2026 00:00:00 +0000

XBOW published that researchers plus AI agents (OpenAI GPT-5.6 Sol) built a three-bug Chrome→Windows full-chain in six days (Aug 2–7), winning Google's first $250,000 Chrome full-chain VRP under 2026 rules. Chain crossed V8 sandbox and renderer sandbox; bugs were reported, fixed, and never used against users. XBOW estimates ~$40k API compute. Caveat: this is a responsible-disclosure research write-up, not an in-the-wild campaign — keep Chrome updated and relaunch after updates.

Defender for Office 365: prompt-injection protection quarantines high-confidence email before Copilot sees it

Tue, 13 Oct 2026 00:00:00 +0000

Microsoft documents a mail-flow layer that detects prompt-injection content in inbound email before delivery to users or AI assistants. Detections land as High confidence phishing with technology "Prompt injection protection" (MDO Plan 2 / E5-class). Focus: instructions that try to exfiltrate via URL, reveal system prompts, or enumerate write-capable tools — including hidden/white-on-white text. Runtime Copilot safeguards remain a second layer. This is harness/context defense at the email channel, not a new CVE.

Hunt: SEO fake-PuTTY MSI → Node BYOI loader with EtherHiding C2

Tue, 13 Oct 2026 00:00:00 +0000

Unit 42 (Oct 12 TTI): SEO-poisoned "PuTTY" downloads serve Putty-x86_64.msi that drops a signed PuTTY 0.85 decoy plus a Node.js bring-your-own-interpreter loader. C2 URL is resolved via Ethereum eth_call (EtherHiding) and polled every 500 ms as fake static assets; operators push recon, cookie theft, and near-live screen streaming.

// KQL sketch — AppData node + msiexec→curl Node zip
DeviceProcessEvents
| where FileName =~ "node.exe" and FolderPath has @"AppData\Local"
| project Timestamp, DeviceName, FolderPath, ProcessCommandLine
DeviceProcessEvents
| where InitiatingProcessFileName =~ "msiexec.exe"
| where FileName in~ ("curl.exe","cmd.exe") and ProcessCommandLine has "nodejs.org"

Also watch eth_call from non-browser processes to public Ethereum RPCs. Sketch only; validate IOCs from Unit 42 before alerting.

Patch Tuesday triage: MotW bypass first, then exposed edge, then RCEs

Tue, 13 Oct 2026 00:00:00 +0000

For October's record Microsoft release, prioritize by exploitation evidence, not CVSS alone: (1) CVE-2026-96951 Windows Shell MotW bypass — already exploited and on CISA KEV (due Oct 16); (2) internet-facing edge already in KEV (Ivanti Connect Secure CVE-2026-104935); (3) high-likelihood RCEs Microsoft flags as "Exploitation More Likely" (Exchange, AD DS, MSMQ, Outlook). Use MSRC risk cohorts for shared-boundary bundles, then confirm with VEX. Restart browsers after Chrome updates so sandbox fixes take effect.

State of AI Cybersecurity: harness-layer attack proposals up 175% (12 → 33), March to October

Tue, 13 Oct 2026 00:00:00 +0000

Rob T. Lee and Indre Altman compared the [un]prompted AI Security Practitioner Conference CFP rounds: 469 submissions in March and 490 in October. Harness-layer attack proposals rose from 12 to 33. 94 submissions used the word 'harness'. 71% of the October pool is agent-native (harness, MCP and skill supply chain, agent identity, prompt injection, agentic-browser attacks). Only 12% carry production evidence. Caveat: this is a CFP census, not incident data. The full report is linked from the conference site; we've only verified the figures quoted in the author's post.

Windows Shell zero-day CVE-2026-96951 exploited; on CISA KEV, due Oct 16

Tue, 13 Oct 2026 00:00:00 +0000

Microsoft's October Patch Tuesday marks this Windows Shell security feature bypass (CVSS 8.8) as 'Exploitation Detected'. CISA added it to KEV on Oct 13 (listed as a Windows command injection) with an Oct 16 federal due date. Patch Windows endpoints first.

Ivanti Connect Secure CVE-2026-104935 added to KEV, due Oct 16

Tue, 13 Oct 2026 00:00:00 +0000

Authentication bypass by spoofing leading to remote command execution (CVSS 8.6). Listed in CISA KEV on Oct 13 with an Oct 16 due date. Apply Ivanti's fix and review appliance integrity.

October 2026 Patch Tuesday: criticals to prioritize after the zero-day

Tue, 13 Oct 2026 00:00:00 +0000

Record-scale Microsoft release (reporting outlets cite ~1,411–1,419 Microsoft CVEs depending on counting). Actively exploited: Windows Shell MotW bypass CVE-2026-96951 (CVSS 8.8, CISA KEV due Oct 16). Publicly disclosed before patch (not known exploited per Microsoft/SANS): .NET Framework WDAC bypass CVE-2026-108398 and MessagePack-CSharp DoS CVE-2026-92707. High-value not-known-exploited: Exchange RCE CVE-2026-70571 (8.1) and AD Domain Services RCE CVE-2026-70229 (8.1), both 'Exploitation More Likely'; Outlook RCEs (e.g. CVE-2026-70242); Message Queuing RCE CVE-2026-100202 (9.8). Severity and exploitability labels are Microsoft's.

REDCap CVE-2026-90817 unauthenticated RCE exploited via public surveys

Tue, 13 Oct 2026 00:00:00 +0000

Unit 42 reports attacks dropping a webshell then a fileless XMRig miner. Fixed in 16.0.49, 17.3.10 and 17.4.4. Not in CISA KEV as of Oct 13; NVD has no entry yet.

VMware vCenter CVE-2026-59310 on CISA KEV since Aug 18 (known ransomware use)

Tue, 13 Oct 2026 00:00:00 +0000

Broadcom VMware vCenter path traversal that lets a network-adjacent attacker execute code. CISA added it to KEV on Aug 18 with an Aug 21 federal due date and marks known ransomware campaign use. A separate dark-web monitor also claimed an exploit for sale (~$10k) on Oct 13; treat that as secondary noise. Patch vCenter and keep it off the public internet.

Agentic Bug Hunter (community bug-bounty toolkit)

Tue, 13 Oct 2026 00:00:00 +0000

Built around Claude Code: recon, known-CVE scanning and attack-surface ranking. Caveats: third-party community project, not audited by us; needs Claude Code or API access and won't run fully on local models without changes; only point it at targets you're authorized to test.

Hunt: AhsayCBS / REDCap post-exploitation (webshell → miner)

Tue, 13 Oct 2026 00:00:00 +0000

Both campaigns end in web-server-spawned shells and miners. Look for the web service process spawning cmd/powershell/sh, new script files in web roots, and outbound connections to mining pools.

// KQL (Defender) sketch
DeviceProcessEvents
| where InitiatingProcessFileName in~ ("java.exe","w3wp.exe","httpd","php-fpm")
| where FileName in~ ("cmd.exe","powershell.exe","sh","bash")
| project Timestamp, DeviceName, InitiatingProcessFileName, ProcessCommandLine

Tune the parent list to your stack. Sketch only; validate before alerting.

Hunt: SonicWall SMA1000 SSRF abuse

Tue, 13 Oct 2026 00:00:00 +0000

Review SMA1000 access logs for unauthenticated requests to unusual Work Place paths, and investigate unexpected connections from the appliance to internal services. Rotate credentials for anything reachable through the appliance's trust context.

Track CISA KEV yourself in one command

Tue, 13 Oct 2026 00:00:00 +0000

KEV is a public JSON feed. Pull the latest additions with:

curl -s https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json | jq '.vulnerabilities | sort_by(.dateAdded) | .[-5:][] | {cveID, dateAdded, dueDate}'

MSRC introduces 'risk cohorts' to help triage October's huge release

Tue, 13 Oct 2026 00:00:00 +0000

Alongside October Patch Tuesday, MSRC published guidance that groups CVEs by shared deployment, exposure or trust-boundary questions, so one assessment covers many CVEs. MSRC is explicit that cohorts don't replace per-CVE assessment or change update guidance: exploitation status and severity still decide priority.

OpenAI: employee arrested for planting a recording device in an executive conference room

Tue, 13 Oct 2026 00:00:00 +0000

OpenAI said it found a hidden recording device, used its own AI to trace it, and referred the case to the FBI. DOJ says Abdulrahman Alfozan was arrested Oct 9 and charged with intercepting oral communications. OpenAI says it has no indication that model weights, source code or customer data were accessed or exfiltrated. The employee was terminated. Insider-threat reminder for AI labs.

Automatic attack disruption needs Defender for Endpoint Plan 2, not Plan 1

Tue, 13 Oct 2026 00:00:00 +0000

A red-teamer reports MDE P2's automatic attack disruption repeatedly cut off an internal test where other products didn't (opinion, one engagement). The licensing fact is from Microsoft Learn: attack disruption needs MDE Plan 2 (or E5-class / Defender for Business). P1 gives prevention and manual response only. Set device groups to Full or Semi automation and keep the Sense agent current.

French supermarket chains hit via refrigeration controllers; food spoilage and prosecutor probe

Mon, 12 Oct 2026 00:00:00 +0000

Attackers remotely forced defrost cycles on vulnerable Danfoss refrigeration controllers (obsolete AK-SM800/AK-SC355 or unpatched AK-SM800A), with reports of programmed temps up to ~50°C. Coopérative U's CEO confirmed impact from Oct 11; Groupement Mousquetaires reported ~150 Intermarché + 30 Netto stores; Carrefour ~80 of 6,000; Casino said ~5% of Monoprix. Paris prosecutor opened a large-scale cyber-attack investigation. OT reminder: keep ICS management interfaces offline or VPN-segmented — Claroty previously found thousands of similar controllers internet-exposed.

AhsayCBS CVE-2026-105133 + CVE-2026-105134 chain exploited since Oct 7

Mon, 12 Oct 2026 00:00:00 +0000

Auth bypass chained with Replication Receiver RCE gives SYSTEM. Huntress observed webshells and a disguised crypto miner. The first fix was incomplete: install hotfix 10.3.4.45 (v10) or 9.15.2.70 (v7-v9), reboot, then hunt for compromise. Not in CISA KEV as of Oct 13.

Bricksforge WordPress plugin CVE-2026-85097 unauthenticated RCE

Mon, 12 Oct 2026 00:00:00 +0000

Patchstack reports first exploitation on Oct 7. Update the plugin or remove it. We could not reach a primary advisory page today, so treat details as preliminary. Not in KEV as of Oct 13.

GitHub Security Lab Fuzzing Taskflow

Sun, 11 Oct 2026 00:00:00 +0000

Open-source taskflows for the seclab-taskflow-agent framework that use an LLM to plan, build and triage fuzzing harnesses. Caveats: needs model API access and real compute; LLM-written harnesses still need human review, and crashes are not vulnerabilities until triaged.

SonicWall SMA1000 CVE-2026-102255 pre-auth SSRF under attack

Fri, 09 Oct 2026 00:00:00 +0000

Pre-auth SSRF in the Work Place interface (advisory SNWLID-2026-0017). Upgrade to 12.4.3-03670 or 12.5.0-03082; note 12.5.0-02952, the September zero-day fix, is now listed as affected. Canada's Cyber Centre cites open-source reports of exploitation. Not in KEV as of Oct 13.

DOJ/FBI disrupt DPRK remote IT worker "laptop farms" affecting 70+ U.S. companies

Fri, 02 Oct 2026 00:00:00 +0000

DOJ announced coordinated actions against North Korean remote IT-worker schemes that place fraudulently hired workers inside U.S. firms via stolen identities and U.S.-based laptop farms. Actions include indictments, arrests of U.S. facilitators, and searches; CrowdStrike published assistance on disrupting FAMOUS CHOLLIMA operations. Defender takeaway: identity proofing for remote hires, geolocation/device anomalies on corporate endpoints, and vendor/contractor access reviews remain high priority.

Microsoft now publishes VEX statements for all Microsoft-assigned CVEs

Tue, 08 Sep 2026 00:00:00 +0000

Since Sept 8, 2026, MSRC has published machine-readable CSAF VEX documents for every Microsoft CVE, extending the Azure Linux pilot from Oct 2025. Feed your scanner from msrc.microsoft.com/csaf/vex/ to cut false positives. It doesn't add any updates you have to deploy.

SafeHarness: defenses built into the agent harness lifecycle

Wed, 15 Apr 2026 00:00:00 +0000

An arXiv paper treating the harness (tool use, context management, state) as the attack surface. It proposes four defense layers: adversarial context filtering, causal verification of decisions, privilege-separated tool control and safe rollback. It reports ~42% lower attack success on benchmarks. Caveat: simulated evaluation, not production evidence.