Week in review: LLM package hallucinations harm supply chains, Nagios Log Server flaws fixed
Vulnerabilities mentionedAll →
| CVE | Vulnerability | CVSS | EPSS | Flags | Affected | Exposure | Published |
|---|---|---|---|---|---|---|---|
| CVE-2021-20035 | Authenticated OS Command Injection in SonicWall SMA100 Appliances CVE-2021-20035 is an OS command injection flaw (CWE-78) caused by improper neutralization of special elements in the management interface of SonicWall SMA100 series appliances. A remote attacker who has authenticated with low-level privileges can inject arbitrary operating system commands, which are executed on the appliance as the 'nobody' user. Per the CVSS scoring, the primary impact is on availability, potentially leading to denial of service, though command execution on the appliance could facilitate further abuse. The flaw affects SMA 200, 210, 400, 410, and 500v firmware. CISA added it to the Known Exploited Vulnerabilities catalog on 2025-04-16, and related reporting describes ongoing attacks against SonicWall SMA 100 devices — including by threat group UNC6148 deploying the OVERSTEP rootkit and tailored backdoor malware, some against fully patched appliances — so defenders should treat this as actively exploited. Do: Patch to the fixed firmware release specified in the SonicWall advisory for your SMA model as soon as possible; federal agencies must follow the BOD 22-01 mitigation deadline per the CISA KEV required action. Restrict the management interface to trusted networks, enforce MFA on the portal, and hunt for signs of compromise such as the OVERSTEP rootkit, unexpected persistence, or unfamiliar accounts, since reporting indicates tailored backdoor malware in recent SMA 100 attacks. | 6.5 | 4% | KEV |
| large≈ tens of thousands of internet-exposed SMA 100-series appliances (order of magnitude 10k–100k) | |
| CVE-2025-24054 | NTLM Hash Disclosure Spoofing Vulnerability in Microsoft Windows CVE-2025-24054 is a spoofing vulnerability in Microsoft Windows NTLM caused by external control of a file name or path (CWE-73): when Windows processes a file whose name or path points to an attacker-controlled resource, the system is induced to authenticate via NTLM to that destination, disclosing the user's NTLM hash. An unauthorized network attacker triggers the flaw by convincing a user to interact with a crafted file, for example downloading or opening a malicious file, which sends NTLM credentials to a host of the attacker's choosing. With the captured NTLM hash, the attacker can attempt relay or offline cracking to spoof and impersonate the user on the network; no privileges are required, but user interaction is needed (CVSS 3.1: 5.4, medium). Any unpatched Windows 10 (1507 through 22H2), Windows 11 (22H2 through 24H2), or Windows Server 2008, 2012, 2016, or 2019 system in environments that use NTLM authentication is affected. The flaw is being actively exploited: CISA added it to the KEV catalog on 2025-04-17, EPSS puts the 30-day exploitation probability at about 59% (99th percentile), public proof-of-concept exploits exist, and multiple attack campaigns have been reported, including targeting of organizations in Poland and Romania shortly after the patch became available. Do: Apply Microsoft's April 2025 security updates across all listed Windows 10, Windows 11, and Windows Server releases and confirm updated builds through patch-management reporting; the flaw was added to CISA's KEV on 2025-04-17, so federal agencies must patch per BOD 22-01 deadlines or apply vendor mitigations. As interim mitigation, restrict outbound NTLM authentication (for example by blocking outbound TCP 445/139 to untrusted hosts, enforcing SMB signing, or removing NTLM where feasible) and hunt for anomalous outbound NTLM authentication following user file downloads or opening of untrusted shortcut files. | 5.4 | 59% | KEV PoC ×3 |
| masshundreds of millions of Windows devices (desktops and servers on the affected Windows 10, Windows 11, and Windows Server releases with NTLM in use) | |
| CVE-2025-31200 +1 in the same advisory: …31201 | Memory Corruption in Apple iOS, iPadOS, macOS Audio Processing Enables Code Execution CVE-2025-31200 is a memory corruption flaw (CWE-119) in Apple's audio stream handling, fixed with improved bounds checking, that allows code execution when a device processes an audio stream in a maliciously crafted media file. An attacker who can deliver such a file to a vulnerable Apple device can gain arbitrary code execution with full confidentiality, integrity, and availability impact (CVSS 3.1: 9.8 critical, network vector). Affected products are iOS, iPadOS, macOS (Sequoia), tvOS, visionOS, and watchOS on versions released before the April 2025 fixes. Apple stated the issue was exploited in an extremely sophisticated attack against specific targeted individuals on iOS versions before 18.4.1, and CISA added it to the KEV on 2025-04-17; EPSS estimates a 19.7% (97th percentile) probability of exploitation within 30 days. Public analyses describe it chained with the WebKit flaw CVE-2025-31201, which Apple patched in the same emergency updates. Do: Update all Apple devices immediately: iOS/iPadOS 18.4.1 or later, macOS Sequoia 15.4.1 or later, tvOS 18.4.1 or later, visionOS 2.4.1 or later, and watchOS 11.5 or later; the same updates also fix the related actively exploited WebKit zero-day CVE-2025-31201. The flaw is in CISA KEV (added 2025-04-17), so US federal agencies must apply the updates per BOD 22-01 or discontinue use. Verify installed OS versions in Settings > General > About (iOS/iPadOS) or About This Mac, and prioritize high-risk/targeted users for immediate patching and review. | 9.8 | 19% | KEV PoC ×2 |
| masswell over 1 billion Apple devices (iOS/iPadOS/macOS active installed base; all devices on pre-18.4.1/15.4.1 OS versions at disclosure were affected) |
Full article1,067 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:
Apple plugs zero-day holes used in targeted iPhone attacks (CVE-2025-31200, CVE-2025-31201)
Apple has released emergency security updates for iOS/iPadOS, macOS, tvOS and visionOS that fix two zero-day vulnerabilities (CVE-2025-31200, CVE-2025-31201) that have been exploited “in an extremely sophisticated attack against specific targeted individuals on iOS.”
When companies merge, so do their cyber threats
For CISOs, mergers and acquisitions (M&A) bring both potential and risk. These deals can drive growth, but they also open the door to serious cybersecurity threats that may derail the transaction.
Windows NTLM vulnerability exploited in multiple attack campaigns (CVE-2025-24054)
CVE-2025-24054, a Windows NTLM hash disclosure vulnerability that Microsoft has issued patches for last month, has been leveraged by threat actors in campaigns targeting government and private institutions in Poland and Romania.
The quiet data breach hiding in AI workflows
As AI becomes embedded in daily business workflows, the risk of data exposure increases. Prompt leaks are not rare exceptions. They are a natural outcome of how employees use large language models. CISOs cannot treat this as a secondary concern.
Cozy Bear targets EU diplomats with wine-tasting invites (again)
APT29 (aka Cozy Bear, aka Midnight Blizzard) is, once again, targeting European diplomats with fake invitations to wine-tasting events, Check Point researchers have shared.
Securing digital products under the Cyber Resilience Act
In this Help Net Security interview, Dr. Dag Flachet, co-founder at Codific, explains what the Cyber Resilience Act (CRA) means for companies and how it compares to GDPR in terms of regulatory complexity and impact on organizations.
Sonicwall SMA100 vulnerability exploited by attackers (CVE-2021-20035)
CVE-2021-20035, an old vulnerability affecting Sonicwall Secure Mobile Access (SMA) 100 series appliances, is being exploited by attackers.
Funding uncertainty may spell the end of MITRE’s CVE program
The future of the Common Vulnerabilities and Exposures (CVE) program hangs in the balance: MITRE, the not-for-profit US organization that runs it, could lose the US federal funding that helps them maintain it.
When AI agents go rogue, the fallout hits the enterprise
In this Help Net Security interview, Jason Lord, CTO at AutoRABIT, discusses the cybersecurity risks posed by AI agents integrated into real-world systems.
Hertz data breach: Customers in US, EU, UK, Australia and Canada affected
American car rental company Hertz has suffered a data breach linked to last year’s exploitation of Cleo zero-day vulnerabilities by a ransomware gang.
Cybercriminal groups embrace corporate structures to scale, sustain operations
In this Help Net Security interview, Sandy Kronenberg, CEO of Netarx, discusses how cybercriminal groups are adopting corporate structures and employee incentives to scale operations, retain talent, and evade detection.
Critical flaws fixed in Nagios Log Server
The Nagios Security Team has fixed three critical vulnerabilities affecting popular enterprise log management and analysis platform Nagios Log Server.
Review: Hands-On Industrial Internet of Things
Hands-On Industrial Internet of Things is a practical guide designed specifically for professionals building and securing industrial IoT (IIoT) systems.
Package hallucination: LLMs may deliver malicious code to careless devs
LLMs’ tendency to “hallucinate” code packages that don’t exist could become the basis for a new type of supply chain attack dubbed “slopsquatting” (courtesy of Seth Larson, Security Developer-in-Residence at the Python Software Foundation).
Tirreno: Open-source fraud prevention platform
Tirreno is an open-source fraud prevention platform designed as a universal analytics tool to monitor online platforms, web applications, SaaS products, digital communities, mobile apps, intranets, and e-commerce websites.
The UK’s phone theft crisis is a wake-up call for digital security
Phone theft is now commonplace in London. The Met Police recently revealed that it seizes 1,000 stolen phones weekly as it cracks down on organized criminal networks driving the £50 million trade. Nationally, cases have doubled to 83,900 annually.
Inside PlugValley: How this AI vishing-as-a-service group operates
In this Help Net Security video, Alexis Ober, Threat Intel Analyst at Fortra, discusses the threat actor group PlugValley, which is now offering AI-powered vishing-as-a-service.
Sector by sector: How data breaches are wrecking bottom lines
Data breaches are rising across industries, hitting healthcare, finance, and retail especially hard. The damage goes beyond lost data, as it’s financial, operational, and reputational.
Strategic AI readiness for cybersecurity: From hype to reality
AI readiness in cybersecurity involves more than just possessing the latest tools and technologies; it is a strategic necessity.
Chief Legal Officers step up in cybersecurity oversight
In this Help Net Security video, Jennifer Chen, Executive Director of the Association of Corporate Counsel (ACC) Foundation, discusses how globally, Chief Legal Officers (CLOs) are becoming integral leaders in cybersecurity strategy, holding leadership positions, and frequently reporting cybersecurity strategies to the company board.
The future of authentication: Why passwordless is the way forward
By now, most CISOs agree: passwords are the weakest link in the authentication chain. They’re easy to guess, hard to manage, and constantly reused.
Why shorter SSL/TLS certificate lifespans matter
Digital certificates are the unsung heroes of the internet, silently verifying that the websites, apps, and services you use are legit and your data is safe. For years, we’ve leaned on certificates with maximum validity term stretching for months and, in some cases, even years.
94% of firms say pentesting is essential, but few are doing it right
Organizations are fixing less than half of all exploitable vulnerabilities, with just 21% of GenAI app flaws being resolved, according to Cobalt.
Browser extensions make nearly every employee a potential attack vector
Despite being present on virtually every employee’s browser, extensions are rarely monitored by security teams or controlled by IT, according to LayerX.
Organizations can’t afford to be non-compliant
Non-compliance can cost organizations 2.71 times more than maintaining compliance programs, according to Secureframe.
Cybersecurity jobs available right now: April 15, 2025
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.
Cyber threats against energy sector surge as global tensions mount
Cyberattacks targeting the energy sector are increasing, driven by a host of geopolitical and technological factors.
SafeLine Bot Management: Self-hosted alternative to Cloudflare
SafeLine is an open-source, self-hosted Web Application Firewall (WAF) and reverse proxy designed for organizations that value autonomy, visibility, and customizability.
New infosec products of the week: April 18, 2025
Here’s a look at the most interesting products from the past week, featuring releases from Cato Networks, Cyware, Entrust, PlexTrac, and Seemplicity.
Text extracted automatically; images, tables and formatting may be missing. Original: https://www.helpnetsecurity.com/2025/04/20/week-in-review-llm-package-hallucinations-harm-supply-chains-nagios-log-server-flaws-fixed/