Week in review: Cisco patches SD-WAN 0-day, unpatched Microsoft Exchange Server flaw exploited
Vulnerabilities mentionedAll →
| CVE | Vulnerability | CVSS | EPSS | Flags | Affected | Exposure | Published |
|---|---|---|---|---|---|---|---|
| CVE-2026-20182 | Authentication Bypass in Cisco Catalyst SD-WAN Control Components CVE-2026-20182 is a critical authentication flaw (CWE-287) in the control-connection peering authentication of Cisco Catalyst SD-WAN Controller (formerly vSmart), Manager (formerly vManage), and Validator (formerly vBond). Because the peering authentication mechanism does not work properly, an unauthenticated, remote attacker can send crafted requests during the control-connection handshake and log in to the controller as an internal, high-privileged, non-root user without valid credentials. With this access, the attacker can reach NETCONF and manipulate network configuration across the entire SD-WAN fabric. Any organization running these Catalyst SD-WAN control components is affected, and the flaw carries a CVSS 3.1 score of 10.0 and a 91.5% EPSS score. CISA added it to the Known Exploited Vulnerabilities catalog on 2026-05-14, confirming exploitation in the wild, amid a series of exploited Cisco SD-WAN zero-days including a compromise at a communications service provider. Do: Upgrade affected Catalyst SD-WAN Controller, Manager, and Validator components to the fixed releases identified in Cisco's May 2026 advisory (version numbers are not provided in this data). Use the advisory's 'show control connections' guidance to inspect control-connection handshaking for anomalies and audit for unauthorized high-privileged non-root accounts and unexpected NETCONF sessions. Operators — especially federal agencies — should follow CISA Emergency Directive 26-03 and CISA's Hunt & Hardening Guidance for Cisco SD-WAN Devices, including restricting internet exposure of SD-WAN management interfaces until patched. | 10.0 | 92% | KEV |
| largeon the order of tens of thousands of affected control-plane systems (controllers, managers, validators) across enterprise, service-provider, and government… | |
| CVE-2026-41940 | Missing-Authentication Bypass in WebPros cPanel & WHM (AuthBypass to RCE) CVE-2026-41940 is a critical missing-authentication flaw (CWE-306) in the login flow of WebPros cPanel & WHM (versions after 11.40) and WP2 (WordPress Squared) that lets unauthenticated remote attackers bypass authentication and gain unauthorized access to the control panel. Because no privileges, user interaction, or special conditions are required, any attacker who can reach the login endpoint over the network can attempt it. Beyond control-panel account takeover, public proofs of concept — including watchTowr's 'AuthBypass to RCE' exploit — show the flaw can be chained to remote code execution, and reporting indicates a single hosting customer could obtain root control of an entire shared server. Any hosting provider, MSP, reseller, or organization running cPanel/WHM or WP Squared is affected; cPanel is the dominant commercial hosting control panel, implying a very large installed base of shared-hosting servers and hosted domains. Exploitation is confirmed in the wild: CISA added it to the KEV on 2026-04-30 with known ransomware use, EPSS assigns a 98.5% probability of exploitation within 30 days (100th percentile), and multiple threat actors are actively exploiting it, including against government and MSP networks. Do: Patch immediately per WebPros' advisory — the source data does not specify fixed version numbers, so follow vendor instructions for exact patched releases; CISA KEV/BOD 22-01 requires federal agencies to apply mitigations or discontinue use by the stated deadline (reported as Sunday). Until patched, restrict access to the cPanel/WHM login interface (IP allowlisting, VPN, or limiting management-interface exposure) and hunt for indicators of compromise such as unexpected control-panel logins, new admin accounts, webshells, or ransomware artifacts. The referenced public PoCs can be used to validate whether your instances are exploitable. | 9.3 | 99% | KEV ransomware PoC ×4 |
| mass≈100,000+ internet-exposed cPanel/WHM servers, spanning tens of millions of hosted domains and millions of end users | |
| CVE-2026-42897 | Cross-Site Scripting in Microsoft Exchange Server Actively Exploited by Laundry Bear CVE-2026-42897 is an improper-neutralization flaw (CWE-79) in Microsoft Exchange Server that lets an unauthenticated remote attacker perform cross-site scripting and carry out spoofing. Per the CVSS vector (AV:N/PR:N/UI:R), exploitation requires a victim to interact with attacker-controlled content — reported attacks by the Russian actor Laundry Bear (TA488) trigger when a crafted email is opened in Exchange's webmail interface (Outlook Web Access). The attacker gains the ability to spoof the victim within their webmail session, and reported intrusions show mailbox access persisting even after organizations rotate credentials. Any organization running on-premises Microsoft Exchange Server or Exchange Server Subscription Edition is potentially exposed, particularly those publishing webmail to the internet; specific affected version ranges are not provided in the available data. Exploitation is confirmed in the wild: the flaw was added to CISA KEV on 2026-05-15 and carries a very high EPSS score of 71.2%, though no public proof-of-concept is known. Do: Apply Microsoft's security update for CVE-2024-42897 to all on-premises Exchange and Exchange Subscription Edition servers per vendor guidance — no specific patch versions are provided in the available data, so consult the vendor advisory for build numbers. Federal agencies must patch or apply mitigations per BOD 22-01 given the KEV listing. Because reported attacks (Laundry Bear/TA488) maintain mailbox access after password resets, treat any suspected compromise as persistent: review OWA access logs and inbox rules for anomalies, and invalidate active webmail sessions and tokens, not just credentials. | 6.1 | 71% | KEV |
| large≈20,000+ internet-exposed Exchange servers (public-scan reporting) | |
| CVE-2026-44413 | In JetBrains TeamCity before 2026.1 2025.11.5 authenticated users could expose server API to unauthorised access In JetBrains TeamCity before 2026.1 2025.11.5 authenticated users could expose server API to unauthorised access NVD description · AI analysis pending | 7.5 | <1% |
| — | ||
| CVE-2026-46300 | In the Linux kernel, the following vulnerability has been resolved: In the Linux kernel, the following vulnerability has been resolved: net: skbuff: preserve shared-frag marker during coalescing skb_try_coalesce() can attach paged frags from @from to @to. If @from has SKBFL_SHARED_FRAG set, the resulting @to skb can contain the same externally-owned or page-cache-backed frags, but the shared-frag marker is currently lost. That breaks the invariant relied on by later in-place writers. In particular, ESP input checks skb_has_shared_frag() before deciding whether an uncloned nonlinear skb can skip skb_cow_data(). If TCP receive coalescing has moved shared frags into an unmarked skb, ESP can see skb_has_shared_frag() as false and decrypt in place over page-cache backed frags. Propagate SKBFL_SHARED_FRAG when skb_try_coalesce() transfers paged frags. The tailroom copy path does not need the marker because it copies bytes into @to's linear data rather than transferring frag descriptors. NVD description · AI analysis pending | 7.8 | 9% |
| — |
Full article2,928 words · extracted from helpnetsecurity.com · click to collapse
Here’s an overview of some of last week’s most interesting news, articles, interviews and videos:
Review: Foundations of Cybersecurity, 2nd edition
Jason Andress has refreshed his introductory security text for No Starch Press. He writes in the introduction that the term security now extends past data center servers to cloud resources, mobile devices, the Internet of Things, and AI. Foundations of Cybersecurity: A Straightforward Introduction book is aimed at newcomers to the field, network and system administrators, and managers who need a working grasp of security concepts.
Rustinel: Open-source endpoint detection for Windows and Linux
Open-source endpoint detection has long been split between Windows-focused tools built around Sysmon and Linux tools built around eBPF or auditd. Defenders running mixed environments have had to stitch together separate pipelines, separate rule sets, and separate maintenance burdens. Rustinel, a Rust-based endpoint agent, is an attempt to collapse that work into a single codebase.
Google researchers uncover criminal zero-day exploit likely built with AI
Google’s threat intelligence researchers have linked a zero-day exploit to AI-assisted development by a criminal group. The exploit targeted a popular open-source web-based system administration tool. It allowed attackers to bypass two-factor authentication once they had valid user credentials. The flaw stemmed from a semantic logic error, a case where a developer hardcoded a trust assumption that contradicted the application’s authentication enforcement.
The hidden smart fridge risks that emerge years after purchase
Household refrigerators are built to last more than a decade. The software, cloud services, and mobile apps that control them are not. A new analysis from Erik Buchmann at Leipzig University maps what happens when those two timelines collide, and the findings reach further than the kitchen.
HEIDI: Free IDE security plugin for open-source vulnerability checks
Open-source dependencies make up a large percentage of the code in production applications, and most vulnerability checks still run late in the pipeline, inside CI/CD systems or after a release ships. Meterian is moving those checks earlier with HEIDI, a free plugin for Visual Studio Code and JetBrains IDEs that flags vulnerable packages and offers one-click upgrades from inside the editor.
Amazon Quick authorization bypass let users reach blocked AI chat agents
Enterprises running Amazon Quick, the AWS business intelligence and agentic AI service, rely on a feature called custom permissions to restrict who inside an account can use AI chat agents. Fog Security founder Jason Kao discovered that those restrictions were enforced only in the user interface for a period earlier this year, and direct calls to the backend API returned successful chat responses from agents that administrators had explicitly disabled.
Researchers open-source a Wi-Fi cyber range for security training
Wireless security training programs lean heavily on generic network labs, with Wi-Fi appearing as a checkbox alongside Bluetooth, Zigbee, and cellular. Hands-on environments dedicated to IEEE 802.11 are uncommon, even as Wi-Fi remains the default on-ramp to corporate networks and a recurring entry point for attackers. A new paper from researchers at the Norwegian University of Science and Technology and the University of the Aegean takes aim at that gap with a cyber range built specifically for Wi-Fi.
Sandyaa: Open-source autonomous security bug hunter
Source code auditing has traditionally relied on static analyzers that flag long lists of potential issues, leaving engineers to sort bugs from noise. A new open-source project from offensive-security firm SecureLayer7 takes a different route, using LLMs to read a codebase, trace how data moves through it, and produce working exploit code for the vulnerabilities it confirms. Their open-source tool, called Sandyaa, was released under an MIT license.
KDE gets over €1 million investment to strengthen security and core infrastructure
European governments and public institutions have been shifting away from proprietary software for years, and the financial infrastructure supporting open-source alternatives is growing to match. Germany’s Sovereign Tech Fund announced today that it is investing more than €1 million in KDE, the open-source project behind the Plasma desktop environment and a broad range of Linux software.
Vector embedding security gap exposes enterprise AI pipelines
Enterprise adoption of retrieval-augmented generation has moved sensitive corporate content into a new storage format that existing security tools cannot inspect. Companies deploying internal AI assistants convert documents into high-dimensional numerical vectors and ship them to embedding services and vector databases over ordinary HTTPS connections. Data loss prevention products scan documents and network traffic, and they read none of it. A research framework called VectorSmuggle, released by Jascha Wanger of ThirdKey under the Apache 2.0 license, demonstrates what an attacker can do with that gap.
Linux developers weigh emergency “killswitch” for vulnerable kernel functions
Linux kernel developers are reviewing a proposal for an emergency risk mitigation mechanism (“Killswitch”) that would allow administrators to disable vulnerable kernel functions at runtime. The proposal, submitted by Linux kernel developer/maintainer Sasha Levin, arrives in the wake of the public disclosure of two privilege escalation vulnerabilities affecting the Linux kernel.
JetBrains TeamCity vulnerability allows privilege escalation, API exposure (CVE-2026-44413)
JetBrains has patched a high-severity vulnerability (CVE-2026-44413) in TeamCity, its popular continuous integration and continuous delivery platform, and is urging organizations with on-premises and self-managed deployments to upgrade to the fixed version or implement a security patch.
Stealthy hackers exploit cPanel flaw in active backdoor campaign (CVE-2026-41940)
Security researchers at XLab have outlined an active attack campaign targeting CVE-2026-41940, the recently disclosed vulnerability in cPanel & WHM, and have linked it to a stealthy hacking group that has been operating largely undetected for years.
Microsoft May 2026 Patch Tuesday: Many fixes, but no zero-days
Microsoft has marked May 2026 Patch Tuesday by releasing fixes for 120+ CVE-numbered vulnerabilities, none of which (for a change) are actively exploited or have been publicly disclosed. Still, some deserve more consideration and should be addressed sooner than others.
Fragnesia: New Linux kernel LPE bug was spawned by Dirty Frag patch (CVE-2026-46300)
Researchers have found and disclosed yet another local privilege escalation (LPE) vulnerability in the Linux kernel: CVE-2026-46300, aka “Fragnesia”. The flaw is in the same class of vulnerabilities as the recently disclosed Dirty Frag bug(s).
Unpatched Microsoft Exchange Server vulnerability exploited (CVE-2026-42897)
A critical cross-site scripting (XSS) vulnerability (CVE-2026-42897) in Microsoft Exchange Server is being exploited by attackers, Microsoft warned on Thursday. A permanent fix is still in the works. In the meantime, Microsoft provided temporary mitigations.
Cisco patches another actively exploited SD-WAN zero-day (CVE-2026-20182)
Cisco has patched yet another Catalyst SD-WAN Controller authentication bypass vulnerability (CVE-2026-20182) that has been exploited as a zero-day by “a highly sophisticated cyber threat actor”.
Closing the AI governance gap in your enterprise
In this Help Net Security video, Casey Bleeker, CEO at SurePath AI, talks about the AI governance gap that exists in almost every organization. Drawing from three years of conversations with IT, business, and security leaders, Casey explains why AI adoption is outpacing governance maturity by a wide margin, creating friction between security teams pushing for responsible use and business leaders worried about falling behind competitors.
The hidden risk of non-human identities in AI adoption
An employee with persistent, unsupervised admin access across critical systems, with no audit trail, no clear owner, and no regular access reviews, would raise immediate concern in most organizations. Yet non-human identities and AI agents are often granted that same kind of persistent, broadly privileged access. As AI adoption grows, that gap is becoming harder to ignore.
Police take down relaunched criminal marketplace with 22,000 users, €3.6 million in revenue
German authorities shut down a relaunched version of the criminal marketplace Crimenetwork and arrested its suspected operator. 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.
Poor security left hackers inside water company network for nearly two years
The UK’s data protection regulator, the Information Commissioner’s Office (ICO), fined South Staffordshire Water’s parent company £963,900 over security failures linked to a cyberattack that exposed the personal data of 633,887 people.
iOS 26.5 is out, bringing encrypted RCS messaging to iPhone and Android users
Apple is bringing long-awaited end-to-end encryption to Rich Communication Services (RCS) messaging between iPhone and Android users in iOS 26.5. The feature is launching in beta for iPhone users running iOS 26.5 on supported carriers and Android users using the latest version of Google Messages.
Six new dnsmasq vulnerabilities open the door to DNS cache poisoning, local root
Recent disclosures have revealed that open-source networking tool dnsmasq is grappling with a serious set of vulnerabilities. The problems span memory safety and input validation, with researchers identifying heap buffer overflows, heap corruption, and code execution bugs among the issues.
Instructure took a risky approach to recover stolen Canvas data
Instructure, the company behind the online learning platform Canvas, said it reached an agreement with the extortion group ShinyHunters to prevent data stolen in a recent breach from being leaked online. According to the company’s website, Canvas has more than 30 million active users worldwide and serves more than 8,000 institutions.
Android pushes new scam, theft, and AI protections in 2026 update wave
Phone scammers spoofing bank caller IDs have driven an estimated $980 million in annual losses worldwide, according to Europol. Android’s 2026 security roadmap takes direct aim at that pattern with a verified call system built in partnership with banks, alongside a wider set of protections covering app behavior, device theft, location data, and on-device AI processing.
Microsoft’s agentic security system found four critical Windows RCE flaws
Microsoft responded to growing competition in AI security by announcing that its new agentic security system helped researchers discover 16 new vulnerabilities in the Windows networking and authentication stack, including four critical remote code execution (RCE) vulnerabilities. Two of the four flaws — CVE-2026-40361 and CVE-2026-40364 — were deemed by Microsoft to be more likely to be exploited.
Signal responds to phishing attacks with new in-app security warnings
Signal is adding new protections for users following recent phishing and social engineering attacks.One of the new protections is an additional warning informing users that profile names on Signal are not verified and can be chosen freely by account holders, making impersonation attempts easier.
WhatsApp adds Incognito Chat for private Meta AI conversations
The company launched Incognito Chat with Meta AI, a feature that lets users hold AI conversations the platform itself cannot read. The rollout will reach WhatsApp and the standalone Meta AI app over the coming months. Incognito Chat runs on top of Meta’s Private Processing technology, the same infrastructure the company introduced earlier for AI tools in WhatsApp.
CERN’s open source KiCad library gives the world 17,000 circuit board components
CERN has released its complete KiCad component library under an open source license, making it available to hardware designers anywhere in the world. The library, maintained by CERN’s Design Office, contains more than 17,000 electronic components in the form of schematic symbols and printed circuit board footprints.
AI cyber capability is speeding past earlier projections
AI cyber capability is improving faster than expected, with newer models surpassing earlier projections, according to the UK government’s AI Security Institute (AISI). AISI measures AI cyber capability using “time horizon benchmarks”, which estimate how long AI systems can complete cybersecurity tasks autonomously compared to human experts.
Microsoft’s WinUI agent plugin trims token use by over 70% during development
Microsoft published a plugin on May 13 that lets GitHub Copilot CLI and Claude Code drive the full WinUI 3 development cycle, from project scaffolding through signed MSIX packaging. The WinUI agent plugin ships one agent, eight skills, and several supporting tools targeting the loop developers run dozens of times a day: scaffold, build, run, test, iterate.
Zombie linkages are keeping expired domains trusted for years
Domains expire, get transferred, and return to the market every day. The systems connected to those domains can continue trusting the original owner long after control has changed. Researchers at USC and the University of Twente examined this problem in three widely used systems: Web PKI, Maven Central, and Ethereum Name Service. They use the term “zombie linkages” to describe lingering trust records that remain active after the original domain owner no longer controls the domain.
Deepfake detection is losing ground to generative models
Deepfake detection has been built around a single question for close to a decade. Given a video or audio clip, is it real or synthetic? Commercial detectors analyze pixels, frequencies, and biometric signals to answer that question, and the best of them post strong accuracy numbers on standard benchmarks. In deployment, performance drops sharply on content from newer generators. Researchers at the Vector Institute think this gap is structural, and closing it means rethinking what the field is trying to detect in the first place.
Rocky Linux launches opt-in security repository for urgent fixes
Rocky Linux has introduced a Security Repository that allows the distribution to ship urgent security fixes ahead of upstream Enterprise Linux when public exploit code exists and upstream patches are unavailable.
Thieves unlock stolen iPhones using cheap tools sold on Telegram
Helping a friend recover a stolen phone, Infoblox researchers uncovered a thriving Telegram-based underground marketplace selling unlocking tools and phishing infrastructure used to monetize stolen iPhones. Activation Lock can remotely disable a stolen iPhone and prevent normal resale, with owners also able to lock individual components.
Google lets Workspace admins apply one policy across all SAML apps
Google has updated Context-Aware Access (CAA) in Google Workspace to introduce a default policy assignment for SAML applications. SAML applications are third-party or internal applications that use the Security Assertion Markup Language (SAML) protocol to enable single sign-on (SSO) with Google Workspace credentials.
Security teams are turning to AI to survive alert overload
The World Economic Forum white paper “Empowering Defenders: AI for Cybersecurity” identified AI as the biggest driver of change in cybersecurity for 94% of survey respondents. The paper found that 77% of organizations already use AI in cybersecurity, with much of the activity focused on phishing detection, anomaly monitoring, vulnerability management and incident response.
The scam economy has found its AI upgrade
Scam attempts continue to reach consumers via email, text messages, social media, online advertising, and phone calls. The volume of exposure has remained stable over the past year, with more than half of consumers encountering scam attempts at least monthly, according to the F-Secure Scam Intelligence & Impacts Report 2026.
Instagram messaging encryption removed, and privacy advocates are pushing back
After introducing optional end-to-end encrypted messaging in 2023, Instagram announced in March 2026 that encryption for direct messages would be discontinued, and the feature was removed on May 8. The change allows Instagram to access direct message content, including images, videos, and voice notes.
OpenAI’s Daybreak uses Codex Security to identify risky attack paths
OpenAI Daybreak is the company’s cybersecurity initiative focused on building AI-assisted software defense into the development process from the start. It combines OpenAI models, Codex Security, and cyber-focused GPT-5.5 variants to help organizations identify, validate, and prioritize software vulnerabilities.
Škoda confirms unauthorized access to its online shop
Car manufacturer Škoda discovered that attackers had exploited a vulnerability in its online shop software and gained temporary unauthorized access to the system. After discovering the incident, the company took the shop offline as a precautionary measure, fixed the vulnerability, referred the incident to a specialized IT forensics team for technical analysis, and reported it to the data protection supervisory authority.
General Motors to pay $12.75 million over driver data sales
General Motors has agreed to a $12.75 million settlement with California over allegations that it unlawfully sold drivers’ location and behavioral data to brokers, marking the largest penalty in the history of the state’s Consumer Privacy Act. Prosecutors say GM made approximately $20 million nationwide from the sales.
Fedora Hummingbird brings the container security model to a Linux host OS
Container image security pipelines have spent the past several years pushing toward minimal footprints, hermetic builds, and continuous CVE remediation. The Fedora Project is now applying that same approach to the host operating system. At Red Hat Summit 2026, Fedora announced Fedora Hummingbird, a container-based rolling Linux distribution delivered as an OCI image.
Machine identities outnumber humans 109 to 1
Organizations manage an average of 109 machine identities for every human identity. AI agents account for a growing share of those identities, with companies expecting AI agent growth of 85% over the next 12 months. Machine identities are projected to increase by 77%, and human identities by 56%, based on data from Palo Alto Networks’ 2026 Identity Security Landscape report.
Microsoft turns Copilot Studio into an AI agent control center
The Microsoft Copilot Studio April 2026 updates improve visibility and governance for admins and expand workflow capabilities for managing agents. Copilot surfaces agent status in the authoring experience, giving admins insight into each agent’s security and protection posture. Customers can identify issues such as authentication gaps or policy impacts and investigate them at the source.
The AI oversight paradox: Is the investment worth the cost of watching it?
Unlike in 2025, when AI adoption and testing drove business strategies, organizations in 2026 want proven ROI before committing budgets, according to a report by Globalization Partners.
Download: The IT and security field guide to AI adoption
Security and IT teams are under pressure to adopt AI, but many are seeing the opposite of what was promised. Tools that demo well don’t hold up in real workflows. Complexity increases. Trust breaks down. And instead of reducing workload, AI can introduce new risks and oversight burdens.
Cybersecurity jobs available right now: May 12, 2026
We’ve scoured the market to bring you a selection of roles that span various skill levels within the cybersecurity field. Check out this weekly selection of cybersecurity jobs available right now.
New infosec products of the week: May 15, 2026
Here’s a look at the most interesting products from the past week Alation, Apricorn, Versa Networks, and TrustCloud.
Text extracted automatically; images, tables and formatting may be missing. Original: https://www.helpnetsecurity.com/2026/05/17/week-in-review-cisco-patches-sd-wan-0-day-unpatched-microsoft-exchange-server-flaw-exploited/