Ollama on DanSec Daily

Cybersecurity Brief – 2026-08-26

Wed, 26 Aug 2026 00:00:00 +0000

Major Incidents or Breaches

Nutex Health, a healthcare and services provider, confirmed a data breach involving unauthorized access and data exfiltration from company servers, with sensitive information exposed^[18]^[31].
The Los Angeles County Museum of Art (LACMA) disclosed that a breach last year exposed customer and employee social security and medical data^[14].
Over 270 Zimbra Collaboration Suite servers have been compromised in ongoing attacks exploiting a high-severity remote code execution vulnerability^[22].
Norway’s shared government digital infrastructure suffered significant disruption due to a large-scale distributed denial-of-service (DDoS) attack, impacting public sector services^[17].

Newly Discovered Vulnerabilities

A critical remote code execution vulnerability in Gitea (CVE-2026-60004) is being actively exploited in the wild. The vulnerability was patched in Gitea version 1.27.1, but attackers are deploying miner-like payloads on unpatched instances^[3]^[32].
Two severe unauthenticated authentication bypass vulnerabilities (CVE-2026-61979 and CVE-2026-15981) in the Xecurify miniOrange SAML 2.0 Single Sign On plugin for WordPress are under active exploitation, allowing attackers to gain admin access^[13]^[35].
A high-severity flaw in Marimo notebook software was patched after it was found to allow attacker-supplied Model Context Protocol (MCP) command execution in specially crafted notebooks^[8].
Researchers disclosed a weakness in NVIDIA NemoClaw and its integration with the Ollama API that enables unauthenticated attackers to take control of local AI agent instances and plant persistent malicious instructions^[6]^[26].

Notable Threat Actor Activity

A commercial phishing-as-a-service toolkit, Mirage2FA, has targeted over 4,500 organisations in the US and EU since 2024, abusing Microsoft 365 login flows to bypass two-factor authentication^[9].
Threat actors are abusing npm and its mirrors to host clusters of malicious packages that redirect users to fake Cloudflare CAPTCHA phishing pages^[10]^[15].
A new campaign uses FTP banners as dead drop resolvers to deliver two novel remote access trojans (E4del and PINHOLE RAT), leveraging FTP banner text as a covert command and control channel^[11].
Iranian cyber actors have been sanctioned by the US government for their role in critical infrastructure breaches^[5].
A newly identified phishing-as-a-service platform, AnonyMousKIT, employs AI voice agents to call Apple device theft victims, attempting to phish for passcodes and two-factor authentication codes to bypass Apple’s Activation Lock^[4]^[16].

Trends, Tools, or Tactics of Interest

Phishing remains the leading initial access vector, responsible for more than half of observed cyberattacks in Q2 2026. Microsoft reported observing 7.6 billion phishing emails during this period^[1]^[2].
Attackers are increasingly targeting identity verification and account recovery processes, rather than just login credentials, to compromise accounts^[19].
Prompt injection attacks are being used to manipulate AI-powered email summarizers and assistants by embedding malicious or hidden instructions in otherwise benign-looking content, including encrypted text^[25]^[41].
AI model poisoning is an emerging concern, with vulnerabilities in local AI agent serving platforms allowing attackers to persistently corrupt models via malicious web content^[6]^[26].
WhatsApp has introduced support for multiple passkeys and stronger two-step verification to enhance account security across iOS and Android^[7]^[21]^[36].
Law enforcement agencies from 22 countries arrested 58 individuals and identified 263 suspects in a global crackdown on cybercrime networks coordinated by African crime groups^[23].

Regulatory or Policy Developments Affecting the Security Industry

The US Department of the Treasury announced new sanctions targeting Iranian cyber actors involved in attacks on critical infrastructure^[5].
Nigeria launched new financing, procurement, and infrastructure policies to support sovereign cloud initiatives aimed at strengthening national cybersecurity and technical capabilities^[24].
The Linux Foundation will govern TRACE, an open standard for AI runtime attestation, with contributions from AMD, Intel, Microsoft, OPAQUE, and TII^[33].
Taiwan charged nine individuals, including staff from Nvidia and Super Micro, for illegal exports of AI servers to China, highlighting ongoing regulatory scrutiny in the AI infrastructure supply chain^[40].

Sources

Hospital operator Nutex Health says data stolen in cyberattack - bleepingcomputer.com
Sensitive Information Exposed in Nutex Health Data Breach - securityweek.com
LACMA data breach last year exposed social security and medical data - bleepingcomputer.com
Hackers breached over 270 Zimbra servers in ongoing attacks - bleepingcomputer.com
Massive DDoS attack disrupts Norway’s government digital services - bleepingcomputer.com
Critical Gitea RCE Actively Exploited as Reported Attack Drops Miner-Like Payload - thehackernews.com
CISA Warns of Exploited Gitea Vulnerability - securityweek.com
Attackers Target miniOrange SAML Flaws That Can Grant WordPress Admin Access - thehackernews.com
WordPress Websites Targeted via MiniOrange Plugin Vulnerabilities - securityweek.com
Marimo Notebook Flaw Could Run MCP Commands Before Cells Execute in Edit Mode - thehackernews.com
A Malicious Webpage Could Poison Your Local AI Model Behind NVIDIA NemoClaw - thehackernews.com
Finding Nemo(Claw): Networking Issue Allows for LLM Poisoning in OpenClaw - darkreading.com
Mirage2FA Surge Hits 4,500 US and EU Companies, Abusing Microsoft 365 Login Flows - thehackernews.com
24 npm Packages Abuse unpkg Mirrors to Host Fake Cloudflare CAPTCHA Pages - thehackernews.com
Hackers abuse npm mirrors to host phishing redirect pages - bleepingcomputer.com
E4del and PINHOLE RATs Turn FTP Banners Into Dead Drops for Malware Commands - thehackernews.com
U.S. Sanctions Iran-Linked Hackers Behind Critical Infrastructure Breaches - thehackernews.com
Fake Apple Support AI Calls Target Stolen-Device Owners for Passcodes and 2FA Codes - thehackernews.com
AnonyMousKIT PhaaS uses voice AI agents to phish iPhone passcodes - bleepingcomputer.com
Report: Phishing Remains the Primary Initial Access Vector - blog.knowbe4.com
CyberheistNews Vol 16 #34 Microsoft Observed 7.6 Billion Phishing Emails in Q2 2026 - blog.knowbe4.com
From Fake Workers to Account Recovery: The Growing Identity Verification Risk - bleepingcomputer.com
Hidden Prompts Trick AI Into False Email Summaries - darkreading.com
Grok fooled into stealing user chat, location data, and more - malwarebytes.com
WhatsApp Adds Multiple Passkeys for Phishing-Resistant Sign-Ins Across iOS and Android - thehackernews.com
WhatsApp adds stronger two-step verification, multiple passkeys - bleepingcomputer.com
WhatsApp Adds Multiple Passkeys and Stronger 2SV in Account Security Update - securityweek.com
Police arrests dozens of suspects in global cybercrime crackdown - bleepingcomputer.com
Nigeria Looks to Sovereign Cloud for Cyber, National Security - darkreading.com
Linux Foundation to Govern TRACE, an Open Standard for AI Runtime Attestation - securityweek.com
Taiwan Charges 9 Over Illegal AI Server Exports to China, Including Nvidia and Super Micro Staff - securityweek.com

Cybersecurity Brief – 2026-05-11

Mon, 11 May 2026 00:00:00 +0000

Major Incidents or Breaches

The Canvas online learning system experienced a cyberattack that disrupted access for tens of thousands of students globally, particularly impacting those studying for final exams. Service has since been restored.
Over 500 organizations across sectors including aviation, critical infrastructure, energy, logistics, public administration, and technology have been targeted in a years-long phishing campaign.
German authorities have shut down the rebooted Crimenetwork criminal marketplace, which had over 22,000 users and more than 100 sellers, and arrested its administrator.

Newly Discovered Vulnerabilities

A critical out-of-bounds read vulnerability has been disclosed in Ollama, allowing remote, unauthenticated attackers to leak the application’s entire process memory.
A new Linux vulnerability, referred to as “Dirty Frag” (also known as Copy Fail 2, CVE-2026-43284 and CVE-2026-43500), was disclosed prior to a patch being available and may already be under exploitation in the wild.

Notable Threat Actor Activity

A malicious repository impersonating OpenAI’s Privacy Filter model reached the top spot on Hugging Face, delivering a Rust-based information stealer to Windows users and amassing 244,000 downloads.
The TrickMo Android banking malware, in campaigns targeting users across Europe, has adopted The Open Network (TON) blockchain for covert command-and-control communications and introduced new functionality.
Attackers are leveraging Google Ads and shared Claude.ai chat links in an active malvertising campaign to distribute Mac malware to users searching for “Claude mac download.”
The Checkmarx Jenkins AST Plugin was compromised in a supply chain attack, with a malicious version published to the Jenkins Marketplace.

Trends, Tools, or Tactics of Interest

Ransomware activity in Q1 2026 shows continued consolidation, with over 2,100 new victims listed across more than 70 active data leak sites.
YARA-X version 1.16.0 has been released, providing four improvements and four bug fixes to the malware identification tool.
Threat actors are increasingly abusing trusted platforms (e.g., Hugging Face, Google Ads, Claude.ai, Jenkins Marketplace) for malware delivery and supply chain attacks.

Regulatory or Policy Developments Affecting the Security Industry

German law enforcement action against the Crimenetwork marketplace demonstrates ongoing international efforts to disrupt cybercriminal infrastructure and prosecute operators.

Cybersecurity Brief – 2026-05-06

Wed, 06 May 2026 00:00:00 +0000

Major Incidents or Breaches

DAEMON Tools supply chain attack: Installers downloaded from the official DAEMON Tools website since 8 April were trojanised, delivering a sophisticated backdoor to thousands of systems globally. Government and scientific entities were among those targeted, though the backdoor was selectively deployed on a limited number of systems.
Vimeo data breach: The ShinyHunters extortion group accessed and exfiltrated personal information of over 119,000 users following a breach of the Vimeo platform in April.
Instructure breach: A hacker claims to have stolen 280 million data records from 8,809 educational institutions by breaching Instructure, exposing student and staff data.
Trellix source code breach: Attackers accessed source code for Trellix security products, potentially exposing detection mechanisms and product controls.
Taiwan high-speed rail hack: A university student was arrested for exploiting the TETRA communication system to trigger emergency brakes on Taiwan’s high-speed railway.
CloudZ RAT exploitation via Windows Phone Link: Researchers disclosed an incident involving the CloudZ remote access tool, leveraging a previously undocumented plugin (“Pheno”) to steal credentials and OTPs.
Government targeting by UAT-8302: China-linked APT group UAT-8302 was attributed to ongoing attacks against government entities in South America and South Asia, using malware shared across regions.

Newly Discovered Vulnerabilities

Palo Alto Networks PAN-OS (CVE-2026-0300): A critical buffer overflow vulnerability in PAN-OS User-ID Authentication Portal is under active exploitation, enabling remote code execution on PA and VM series firewalls.
Apache HTTP/2 (CVE-2026-23918): A severe vulnerability in Apache HTTP Server’s HTTP/2 implementation could allow denial-of-service and potential remote code execution. Security updates have been released.
MetInfo CMS (CVE-2026-29014): A critical flaw in the open-source MetInfo CMS is being actively exploited for remote code execution attacks.
Android System (CVE-2026-0073): A critical remote code execution vulnerability was patched in Android’s System component, exploitable without user interaction.
Ollama “Bleeding Llama”: A critical heap out-of-bounds read issue in Ollama LLM deployments could be exploited remotely and without authentication, potentially affecting up to 300,000 deployments.
WhatsApp: Two newly patched vulnerabilities could have allowed attackers to send malicious files to users, potentially leading to malware infection.
Microsoft Edge: Proof-of-concept exploits show that passwords are stored in process memory and can be extracted by users with admin privileges, posing an enterprise risk.

Notable Threat Actor Activity

ShinyHunters: Responsible for the Vimeo data breach, exfiltrating user personal information.
UAT-8302 (China-linked APT): Conducting cross-regional campaigns targeting government agencies using shared APT malware.
Karakurt ransomware group: A Latvian national, Deniss Zolotarjovs, was sentenced in the US for his role as a negotiator in extortion operations.
Quasar Linux (QLNX) malware: A new, previously undocumented Linux implant is targeting software developers with rootkit, backdoor, and credential-stealing features.

Trends, Tools, or Tactics of Interest

Supply chain attacks: DAEMON Tools and Trellix incidents highlight the ongoing risk of supply chain compromise, with attackers leveraging trusted software distribution channels.
OAuth token persistence: Many AI tools and productivity apps connected to Google or Microsoft environments leave persistent OAuth tokens without expiration or automated revocation, increasing lateral movement and persistence risks.
AI service exposure: Scans of one million exposed AI services reveal significant security gaps as businesses rapidly adopt AI technologies, with insecure deployments and insufficient access controls.
Social engineering and AI: Evolution in phishing and social engineering tactics, now leveraging AI to create more convincing lures and automate attacks, as noted in new KnowBe4 SAT platform updates and recent phishing campaigns.
Binary Transparency for Android: Google has expanded public verification of Android app binaries to prevent supply chain attacks.
Microsoft phishing campaign: Ongoing sophisticated phishing attacks against US organisations, using adversary-in-the-middle (AitM) techniques to harvest credentials.
Jailbreaking and data poisoning AI models: Security researchers are actively probing AI guardrails for weaknesses, using jailbreaking and data poisoning to demonstrate attack vectors.

Regulatory or Policy Developments

FTC action against Kochava: The US Federal Trade Commission will ban data broker Kochava and its subsidiary from selling Americans’ location data without explicit consent, as part of a legal settlement.
Oracle patch cadence: Oracle will now release monthly critical security patch updates, aiming to address high-priority vulnerabilities more rapidly.
SSL.com root certificate rotation: SSL.com rotated its root certificate as part of standard lifecycle management.
CISA critical infrastructure initiative: The US Cybersecurity and Infrastructure Security Agency (CISA) announced a new initiative to strengthen the security of critical infrastructure.

Cybersecurity Brief – 2026-01-31

Sat, 31 Jan 2026 00:00:00 +0000

Major Incidents or Breaches

CERT Polska reported coordinated cyber attacks on over 30 wind and solar farms, as well as a private manufacturing company, in Poland. Related reporting attributes a Russia-linked threat actor (Sandworm/Electrum) as responsible for these intrusions, which resulted in bricked ICS devices at 30 sites affecting communication and control systems in the Polish power grid.
ShinyHunters ransomware group claimed breaches of Match Group dating platforms (Match, Hinge, OkCupid) and Panera Bread, allegedly stealing millions of user records.
Illegal cryptocurrency flows reached $158 billion in 2025, marking a record high after several years of decline.
Researchers identified malicious Google Chrome extensions capable of hijacking affiliate links, stealing data, and collecting OpenAI ChatGPT authentication tokens.
Android users were targeted with a remote access trojan (RAT) delivered via malicious applications hosted on Hugging Face repositories.
Over 175,000 exposed Ollama AI model hosts were identified, with 23,000 persistently active, posing risks of LLM abuse.

Newly Discovered Vulnerabilities

Ivanti released patches for critical-severity, actively exploited zero-day vulnerabilities in its Endpoint Manager Mobile (EPMM) product, which allowed unauthenticated remote code execution.
Microsoft fixed a bug in Microsoft 365 Outlook that prevented users from accessing encrypted emails in classic Outlook.
The Windows 11 KB5074105 update addressed multiple issues, including boot, sign-in, and activation failures.

Notable Threat Actor Activity

China-linked threat actor UAT-8099 was observed targeting IIS servers in Asia with the BadIIS SEO malware campaign between late 2025 and early 2026. The campaign focused on search engine manipulation through server compromise.
Sandworm/Electrum, a Russia-linked group, was attributed to the bricking of ICS devices during coordinated attacks on Polish energy infrastructure.

Trends, Tools, or Tactics of Interest

Agentic AI and autonomous AI assistants such as OpenClaw (ClawdBot, MoltBot) are increasingly present in business environments, raising concerns about privileged, autonomous control and the expanding attack surface.
Security researchers noted increased abuse of Google Presentations for phishing campaigns targeting webmail users.
The Tenable One AI Exposure add-on was introduced to discover unsanctioned AI use within organizations and enforce compliance with approved AI tools.
Aisy launched an AI-assisted vulnerability management platform after emerging from stealth with $2.3 million in seed funding.

Regulatory or Policy Developments

Microsoft announced plans to disable the legacy NTLM authentication protocol by default in future Windows releases, citing security vulnerabilities.
The White House revoked two Biden-era memorandums related to software security rules, although some associated resources remain available for government use.
TikTok updated its privacy policy to explicitly reference immigration status in user data, drawing public attention and scrutiny.

Cybersecurity Brief – 2026-01-30

Fri, 30 Jan 2026 00:00:00 +0000

Major Incidents or Breaches

Match Group, the parent company of Tinder, OkCupid, Hinge, and Match.com, confirmed a cybersecurity incident resulting in the compromise of user data across multiple platforms.
Marquis Software Solutions reported that a ransomware attack in August 2025 affecting dozens of U.S. banks and credit unions was due to a breach of SonicWall cloud backups.
An AI chat toy company, Bondu, exposed over 50,000 logs of conversations between children and its AI-enabled toys due to an unsecured web console.
The French national unemployment agency was fined €5 million for a breach that resulted in the theft of job seekers’ personal information.
eScan antivirus suffered a supply chain attack, with malicious updates detected and distributed through its update mechanism.
Google, in collaboration with industry partners, disrupted the IPIDEA residential proxy network, which had been leveraged by threat actors via malware-infected devices.

Newly Discovered Vulnerabilities

Ivanti released security updates for two zero-day remote code execution (RCE) vulnerabilities in Ivanti Endpoint Manager Mobile (EPMM), CVE-2026-1281 and CVE-2026-1340, both of which have been actively exploited.
SolarWinds patched four critical vulnerabilities in its Web Help Desk product, including unauthenticated RCE and authentication bypass flaws.
Multiple critical vulnerabilities were identified in the n8n automation platform, allowing unauthenticated remote code execution and server hijacking via sandbox and AST sanitization flaws.
Microsoft issued an emergency patch for a Microsoft Office zero-day that allowed malicious documents to bypass security checks.
Researchers found 175,000 publicly exposed Ollama AI servers across 130 countries, creating a significant attack surface for exploitation.
Android theft protection features were updated by Google to strengthen device authentication and recovery.

Notable Threat Actor Activity

A new Android malware campaign abused the Hugging Face platform to distribute thousands of APK payload variants targeting financial and payment credentials.
Chinese APT groups deployed advanced malware against organisations in Asia, indicating ongoing high-end cyber-espionage operations.
The Aisuru (Kimwolf) botnet executed a record-setting DDoS attack in December 2025, peaking at 31.4 Tbps and 200 million requests per second.
Flare researchers highlighted how infostealer malware is being distributed via malicious Roblox game mods, enabling lateral movement from home PCs to corporate environments.
An operation dubbed “Bizarre Bazaar” targeted exposed large language models (LLMs) and managed control panels (MCPs) for commercial monetization.
The Clawdbot malware rebranded as Moltbot, sparking impersonation and supply chain risk campaigns.

Trends, Tools, or Tactics of Interest

SentinelOne and Censys reported a widespread issue of unmanaged, publicly accessible AI deployments, specifically Ollama servers, posing risks to data privacy and security.
The use of “semantic chaining” in prompt engineering was shown to bypass content controls in LLMs such as Gemini Nano Banana and Grok 4 by splitting malicious prompts into discrete segments.
The IPIDEA proxy network relied on SDKs embedded in mobile and desktop applications to enroll compromised devices into its proxy service.
The Global Cybersecurity Outlook 2026 and industry commentary emphasize the need for improved vulnerability management, preparation for quantum threats, and responsible AI deployment.
A survey of over 100 energy sector OT systems revealed pervasive cybersecurity gaps in substations, power plants, and control centers.
Microsoft Teams will introduce a feature to report suspicious calls as potential scams or phishing attempts by mid-March 2026.

Regulatory or Policy Developments

The Trump administration rescinded Biden-era guidance requiring U.S. federal agencies to solicit software bills of materials (SBOMs) and compliance attestations with NIST’s Secure Software Development Framework (SSDF) from vendors.
The French data protection authority imposed a €5 million fine on the national unemployment agency for inadequate security measures leading to a data breach.

Cybersecurity Brief – 2025-11-08

Sat, 08 Nov 2025 00:00:00 +0000

Major Incidents or Breaches

The U.S. Congressional Budget Office confirmed it was hacked, with potential exposure of sensitive government data.
Multiple Russian state-sponsored groups expanded destructive cyberattacks against Ukrainian entities, now targeting the grain sector and associated European organisations.
18 individuals were arrested in connection with international credit card fraud rings, responsible for defrauding 4.3 million cardholders in 193 countries of approximately €300 million between 2016 and 2021.

Newly Discovered Vulnerabilities

QNAP patched seven zero-day vulnerabilities in its NAS devices, which were exploited during the Pwn2Own Ireland 2025 competition.
Samsung Galaxy Android devices were affected by a now-patched zero-day vulnerability (CVE-2025-21042) in the image processing library, exploited via malicious WhatsApp images to deliver the LANDFALL spyware.
Cisco reported that two previously exploited zero-day vulnerabilities in ASA and FTD firewalls are now being actively abused to trigger denial-of-service (reboot loop) conditions.
Google released Chrome 142, addressing a high-severity out-of-bounds write vulnerability in WebGPU (CVE-2025-12725) that could permit remote code execution.
The Keras deep learning tool was found to contain a data exposure vulnerability (CVE-2025-12058), allowing arbitrary file loading and server-side request forgery (SSRF) attacks.
Security researchers identified vulnerabilities in AI infrastructure products Ollama and Nvidia, including one that enables remote code execution.

Notable Threat Actor Activity

A China-linked threat actor targeted a U.S. non-profit using legacy vulnerabilities such as Log4j and IIS for long-term espionage and persistence, as part of broader campaigns against U.S. entities.
The LANDFALL Android spyware was deployed in targeted attacks against Samsung Galaxy users in the Middle East, exploiting a zero-day to access device cameras, microphones, location, and data.
Russian state-sponsored groups escalated cyber operations against Ukraine and its European partners, focusing on destructive attacks in the grain sector.
AI-fueled phishing attacks are surging in Africa, with Microsoft and Group-IB reporting increased sophistication and volume.
Malicious apps impersonating AI tools like ChatGPT and DALL-E are being distributed to deliver malware on mobile devices.
A phishing campaign is using invisible hyphens and other invisible Unicode characters in emails to evade security filters.
The Gootloader malware campaign has resurfaced, continuing to target organisations globally.

Trends, Tools, or Tactics of Interest

Malicious NuGet packages were discovered with hidden logic bombs set to detonate in 2027 and 2028, designed to sabotage databases and Siemens S7 industrial control systems via time-delayed payloads.
Visual Studio Code extensions with ransomware-like features, including data encryption and exfiltration, were published openly, with evidence of AI-generated code.
ClickFix phishing campaigns are evolving, now including embedded tutorial videos and tailored prompts for macOS users to increase infection success rates.
Fake CAPTCHA sites are using instructional videos to guide victims through malware installation.
ID verification laws are driving organisations to store large volumes of sensitive data, increasing the risk and impact of breaches.
AI security agents and synthetic personas are being introduced for security operations centres, raising new governance and risk considerations.
AI agents are being mismanaged due to the misapplication of human identity frameworks, potentially leading to high-speed automated incidents.

Regulatory or Policy Developments Affecting the Security Industry

Google’s $32 billion acquisition of Wiz has cleared DOJ antitrust review, pending further regulatory approvals.
Google launched a new feature on Google Maps allowing businesses to report extortion attempts involving fake negative reviews.
Germany is taking measures against the use of Huawei technology in critical national infrastructure, with additional operational technology security guidance following an F5 hack.
Microsoft announced a partnership with UAE-based G42 to build a 5-gigawatt AI campus, prompting security concerns regarding data handling and geopolitical implications.