Microsoft has announced that it is investigating a significant issue with the traditional Outlook desktop client for the Windows operating system, which causes the application to crash upon startup. When a user tries to open the classic version of Outlook for Windows, the problem occurs. Instead of loading the mailbox, the application displays an error message stating: “Cannot start Microsoft Outlook. Cannot open the Outlook window. Cannot open the folder set. The attempt to log on to Microsoft Exchange failed.” To identify users potentially affected by this specific bug, Microsoft has provided IT administrators and technical teams with a precise diagnostic tool

A new hacker group linked to the Chinese Communist Party has been identified by experts at Palo Alto Networks. Unit 42, the California-based company’s threat intelligence division, has released a report revealing the existence of “Phantom Taurus”, a state-run organization that has been engaged in cyber espionage against government and diplomatic institutions for years. Targeted attacks and strategies consistent with Beijing’s interests According to the document, over the past three years the group has conducted covert operations against foreign ministries, embassies and telecommunications companies in the Middle East, Africa and Asia . The leaked information concerns geopolitical issues, foreign relations, and military

Generative Artificial Intelligence (GenAI) systems are revolutionizing the way we interact with technology, offering extraordinary capabilities in the creation of text, images, and code. However, this innovation brings with it new risks in terms of security and reliability. One of the main emerging risks is Prompt Injection , an attack that aims to manipulate the model’s behavior by exploiting its linguistic abilities. We will explore the phenomenon of Prompt Injection in a chatbot in detail, starting with the basics of prompts and Retrieval-Augmented Generation (RAG) systems, then analyze how these attacks occur and, finally, present some mitigations to reduce the risk, such

A study conducted at the Champalimode Foundation in Portugal has shown that a simple video recording is enough to “peek” into the thought process. Scientists discovered that mice’s facial expressions reflect their internal decision-making strategies. This finding paves the way for noninvasive studies of brain function, but also raises new questions about the boundaries of mental privacy. In previous experiments, rodents were presented with a task involving two bowls of water. Only one of them contained sugar water at a given time, and the animals had to guess where the reward was located. As the source changed, the mice had to adapt

Microsoft is accelerating the in-house production of semiconductors for its data centers, aiming to reduce its dependence on external suppliers like Nvidia and AMD. Kevin Scott, the tech giant’s Chief Technology Officer , said this on Wednesday, emphasizing that the long-term strategy calls for the predominant adoption of chips developed in-house. Data centers, the heart of the processing required for developing applications and artificial intelligence models, currently rely largely on GPUs from Nvidia, the undisputed leader in the industry, and to a lesser extent on AMD. However, Microsoft, like other cloud computing players, including Google and Amazon, is investing in the development

Starting September 1, 2025, all artificial intelligence (AI)-generated content in China must be marked with an “electronic watermark,” according to new legislation. The stated goal is to improve the transparency of digital content and reduce the spread of false information. In March this year, the Cyberspace Administration of China (CAC) , together with four other departments, released the “Measures for Identifying Artificial Intelligence-Generated Synthetic Content” (known as the “Identification Measures” ). According to these rules, text, images, audio, video, and virtual scenes created using AI must carry both explicit identifiers, visible to users, and implicit identifiers, embedded in the data that produces

A new phishing toolkit, MatrixPDF , has been discovered that allows attackers to transform ordinary PDF files into interactive decoys that bypass email security and redirect victims to websites that steal credentials or download malware. Varonis researchers, who discovered the tool, note that MatrixPDF is advertised as a phishing simulator and a solution for black team specialists . However, they emphasize that it was first discovered on hacker forums. ” MatrixPDF: An advanced tool for creating realistic phishing PDFs, designed for black-team teams and cybersecurity training,” the announcement reads. “With drag-and-drop PDF import, real-time preview, and customizable overlays, MatrixPDF enables you to

Broadcom has fixed a severe privilege escalation vulnerability in VMware Aria Operations and VMware Tools that was exploited in attacks starting in October 2024. The issue has been assigned the identifier CVE-2025-41244. While the company did not report an exploit in the official bulletin , NVISO researcher Maxime Thibault reported in May that the attacks began in mid-October 2024. Analysis linked the attacks to the Chinese group UNC5174 . The vulnerability allows an unprivileged local user to inject a malicious binary into directories that match generic regular expressions . A variant observed in real-world attacks uses the /tmp/httpd directory. For the malware

A wonderfully widespread argument among those who work with personal data is that of underestimating the risks or refusing to address them at all. This is the belief that there’s no need to worry about processing “non-sensitive” data. The ontological premise for seeking solutions and corrective measures in the areas of lawfulness and security is the ability to ask the right questions. This is why a tendency to overly easily skip data cannot constitute a functional or even minimally useful strategy. Of course, sensitive data exists under the GDPR and requires high levels of protection. However, this doesn’t mean that all other

A team of researchers has developed a simple hardware tool that challenges the fundamental principles of trusted computing in modern cloud environments. Using a device costing less than $50, they were able to bypass the hardware protections of Intel Scalable SGX and AMD SEV-SNP , which enable Trusted Execution Environments (TEEs). These technologies are the foundation of confidential computing used by major cloud providers and protect in-memory data from privileged attacks and physical access, including cold restarts and memory bus interceptions. The developed device is a DDR4 interposer that is inserted between the processor and the memory stick. It manipulates address lines