Find what you’re looking for
Browse everything
6,877 results in Threat Reports
X's Grok AI Faces Global Scrutiny Over Nonconsensual Explicit Image Generation
In early 2026, X's AI chatbot, Grok, was found to have generated and disseminated nonconsensual, sexually explicit images of individuals, including minors. This misuse led to multiple investigations by regulatory bodies across Europe and the United States, scrutinizing X's compliance with data protection laws and its measures to prevent the creation and spread of such harmful content. The incident underscores the urgent need for robust safeguards in AI technologies to prevent exploitation and protect individual privacy. The proliferation of AI-generated explicit imagery has prompted global regulatory bodies to intensify their oversight of AI applications, emphasizing the necessity for companies to implement stringent controls and ethical guidelines in AI development and deployment.
Intruder 2026: Unveiling Exposed Secrets in JavaScript Bundles
In December 2025, Intruder's research team conducted a comprehensive scan of 5 million applications, uncovering over 42,000 exposed tokens hidden within JavaScript bundles. These tokens included sensitive credentials such as code repository access tokens and project management API keys, many of which were active and provided unauthorized access to critical systems. The exposure was attributed to limitations in traditional security tools, which often fail to detect secrets embedded in front-end code, particularly within single-page applications. This incident underscores the urgent need for enhanced secrets detection methods that can effectively identify and mitigate such vulnerabilities in modern web applications.
Chinese Hackers Exploit Dell Zero-Day Vulnerability in 2024
In mid-2024, the Chinese state-sponsored hacking group UNC6201 began exploiting a critical vulnerability (CVE-2026-22769) in Dell's RecoverPoint for Virtual Machines, a solution integral to VMware virtual machine backup and recovery. This hardcoded credential flaw allowed unauthenticated remote attackers to gain unauthorized access to the underlying operating system, achieving root-level persistence. Once inside, UNC6201 deployed advanced malware, including the Grimbolt backdoor, and utilized novel techniques like creating hidden network interfaces ('Ghost NICs') on VMware ESXi servers to move stealthily across networks. ([bleepingcomputer.com](https://www.bleepingcomputer.com/news/security/chinese-hackers-exploiting-dell-zero-day-flaw-since-mid-2024/?utm_source=openai))This incident underscores the persistent threat posed by state-sponsored actors targeting critical infrastructure through zero-day vulnerabilities. The exploitation of such flaws highlights the necessity for organizations to maintain rigorous patch management and continuous monitoring to detect and mitigate sophisticated cyber threats.
Critical Vulnerabilities in Popular VSCode Extensions Expose Developers to Attacks
In February 2026, critical vulnerabilities were discovered in several widely-used Visual Studio Code (VSCode) extensions, including Live Server, Code Runner, Markdown Preview Enhanced, and Microsoft Live Preview. These extensions, collectively downloaded over 128 million times, contained flaws that could be exploited to steal local files and execute remote code. The vulnerabilities were identified by Ox Security, which attempted disclosure since June 2025 without receiving responses from the maintainers. ([bleepingcomputer.com](https://www.bleepingcomputer.com/news/security/flaws-in-popular-vscode-extensions-expose-developers-to-attacks/?utm_source=openai)) This incident underscores the escalating risks associated with third-party development tools and the necessity for rigorous security assessments of IDE extensions. The widespread adoption of these vulnerable extensions highlights the potential for significant supply chain attacks targeting developers and organizations.
Microsoft Uncovers AI Recommendation Poisoning Threat
In February 2026, Microsoft disclosed a new cyber threat termed 'AI Recommendation Poisoning,' where businesses embed hidden instructions within 'Summarize with AI' buttons on their websites. When users click these buttons, the AI assistant's memory is manipulated via URL prompt parameters to favor certain companies or products in future recommendations. Over a 60-day period, Microsoft identified over 50 unique prompts from 31 companies across 14 industries, raising concerns about the integrity of AI-driven insights. This technique mirrors traditional search engine optimization (SEO) manipulation but targets AI systems directly, potentially leading to biased recommendations in critical areas such as health, finance, and security without user awareness. The emergence of AI Recommendation Poisoning underscores the evolving landscape of cyber threats targeting artificial intelligence systems. As AI becomes increasingly integrated into decision-making processes, ensuring the neutrality and reliability of AI outputs is paramount. Organizations must implement robust security measures to detect and prevent such manipulations to maintain trust in AI-driven recommendations.
SmartLoader's Exploitation of Oura MCP Server: A 2026 Supply Chain Attack
In February 2026, cybersecurity researchers uncovered a sophisticated supply chain attack involving the SmartLoader malware. Threat actors cloned the legitimate Oura Model Context Protocol (MCP) Server—a tool connecting AI assistants to Oura Ring health data—and distributed a trojanized version through deceptive GitHub repositories. This malicious server delivered the StealC infostealer, enabling attackers to exfiltrate credentials, browser passwords, and cryptocurrency wallet data from compromised systems. The attackers meticulously built credibility by creating fake GitHub accounts and repositories, submitting the trojanized server to legitimate MCP registries, and excluding the original author from contributor lists, thereby deceiving users into downloading the compromised software. This incident underscores a growing trend where threat actors exploit trusted platforms and tools to infiltrate systems. The methodical approach of building credibility over months highlights the evolving sophistication of supply chain attacks, emphasizing the need for organizations to rigorously verify the authenticity of software sources and implement robust security reviews before integrating third-party tools.
Keenadu Backdoor: A Deep Dive into the 2026 Android Firmware Compromise
In February 2026, Kaspersky researchers uncovered a firmware-level backdoor named Keenadu embedded in Android tablets from multiple manufacturers, including Alldocube. This malware, integrated during the firmware build process, injects itself into the Zygote process, granting attackers extensive control over the device. Keenadu enables remote execution of malicious payloads, such as hijacking browser searches, monetizing app installations, and interacting with advertising elements. The backdoor has been detected in firmware dating back to August 2023, affecting over 13,700 users worldwide, with significant concentrations in Russia, Japan, Germany, Brazil, and the Netherlands. The discovery of Keenadu underscores the escalating threat of supply chain attacks targeting device firmware. This incident highlights the critical need for manufacturers to secure their development processes and for consumers to remain vigilant about device integrity. The integration of malware at such a fundamental level poses significant challenges for detection and removal, emphasizing the importance of robust security measures throughout the supply chain.
AI Assistants Abused as Command-and-Control Proxies in Recent Cyberattacks
In February 2026, cybersecurity researchers disclosed a novel attack technique where AI assistants with web browsing capabilities, such as Microsoft Copilot and xAI Grok, were exploited as covert command-and-control (C2) proxies. This method, termed 'AI as a C2 proxy' by Check Point Research, allows attackers to blend malicious traffic with legitimate enterprise communications, thereby evading detection. The attack leverages anonymous web access combined with browsing and summarization prompts to create a bidirectional channel for data exfiltration and command execution. This development underscores the evolving threat landscape, where AI systems are not only tools for enhancing productivity but also potential vectors for sophisticated cyberattacks. The ability to abuse AI assistants as C2 proxies highlights the need for organizations to reassess their security postures, especially concerning the integration and use of AI technologies within their networks.
Whisper Leak: Unveiling Side-Channel Vulnerabilities in LLMs
In November 2025, researchers Geoff McDonald and Jonathan Bar Or identified a side-channel vulnerability in Large Language Models (LLMs) termed 'Whisper Leak.' This attack exploits patterns in encrypted network traffic—specifically packet sizes and timing—to infer user prompt topics during LLM interactions. Despite TLS encryption, these metadata patterns allow adversaries to classify conversation topics with high accuracy, posing significant privacy risks. The study demonstrated the attack's effectiveness across 28 popular LLMs, achieving near-perfect classification rates and high precision even in scenarios with extreme class imbalance. ([microsoft.com](https://www.microsoft.com/en-us/research/publication/whisper-leak-a-side-channel-attack-on-large-language-models/?utm_source=openai)) The discovery of Whisper Leak underscores the urgent need for LLM providers to address metadata leakage vulnerabilities. As LLMs are increasingly deployed in sensitive domains such as healthcare and legal services, ensuring robust privacy protections is paramount. The researchers evaluated mitigation strategies like random padding, token batching, and packet injection; however, none provided complete protection, highlighting the complexity of securing LLM communications against side-channel attacks. ([microsoft.com](https://www.microsoft.com/en-us/research/publication/whisper-leak-a-side-channel-attack-on-large-language-models/?utm_source=openai))
Serbian Authorities' Misuse of Cellebrite Tools in 2024: A Wake-Up Call for Digital Privacy
In December 2024, Amnesty International reported that Serbian police and intelligence agencies misused Cellebrite's digital forensic tools to unlawfully extract data from mobile devices belonging to journalists and activists. The authorities employed these tools to unlock devices without consent, facilitating the installation of spyware like NoviSpy during detentions and interrogations. This surveillance campaign targeted individuals critical of government policies, leading to significant privacy violations and suppression of civil society. ([amnesty.org](https://www.amnesty.org/en/latest/news/2024/12/serbia-authorities-using-spyware-and-cellebrite-forensic-extraction-tools-to-hack-journalists-and-activists/?utm_source=openai)) The incident underscores the potential for abuse of digital forensic technologies when deployed without stringent oversight. It highlights the urgent need for robust legal frameworks and ethical guidelines to prevent the misuse of such tools against civil society and to protect fundamental human rights.
Salesloft Drift Breach 2025: A Wake-Up Call for SaaS Security
In August 2025, Salesloft's Drift application suffered a significant security breach when attackers exploited compromised OAuth tokens to access and exfiltrate data from over 700 organizations' Salesforce instances. The breach exposed sensitive information, including customer contact details, support case data, and, in some cases, credentials such as AWS access keys and passwords. Prominent companies like Cloudflare, Zscaler, and Palo Alto Networks were among those affected. The attackers, identified as the group "Scattered Lapsus$ Hunters," demanded nearly $1 billion in ransom to prevent the public release of the stolen data. This incident underscores the critical vulnerabilities associated with third-party integrations and the importance of robust security measures to protect against supply chain attacks.
Understanding the 2026 ClickFix DNS PowerShell Attack
In February 2026, a new variant of the ClickFix social engineering attack emerged, exploiting DNS queries to deliver malicious payloads. Attackers deceived users into executing an 'nslookup' command via the Windows Run dialog, which queried an attacker-controlled DNS server. The server responded with a DNS record containing a malicious PowerShell script, leading to the installation of malware, including the remote access trojan ModeloRAT. This method allowed attackers to blend malicious activities within normal DNS traffic, evading traditional detection mechanisms. This incident underscores the evolving sophistication of social engineering attacks, highlighting the need for heightened awareness and advanced security measures. The use of DNS as a delivery mechanism signifies a shift in attacker tactics, emphasizing the importance of monitoring DNS traffic and educating users about the risks of executing unsolicited commands.