New Intel CPU Vulnerabilities Uncovered: Memory Leaks and Spectre v2 Exploits Persist

May 16, 2025
Hardware Security / Vulnerability

Researchers at ETH Zürich have identified a critical new security flaw that affects all modern Intel CPUs, allowing the leakage of sensitive data from memory. This latest vulnerability, dubbed Branch Privilege Injection (BPI), showcases that the Spectre threat continues to impact computer systems over seven years after its initial discovery. According to ETH Zürich, BPI can be exploited to manipulate the CPU’s prediction calculations, granting unauthorized access to information from other users on the same processor. Kaveh Razavi, head of the Computer Security Group (COMSEC) and a co-author of the study, noted that this flaw affects all Intel processors, potentially allowing malicious actors to access the cache contents and working memory of different users sharing the CPU. The attack exploits Branch Predictor Race Conditions (BPRC), which arise when a processor alternates between prediction calculations for multiple users.

New Vulnerabilities in Intel CPUs Expose Risks of Memory Leaks and Spectre Attacks

May 16, 2025

In a significant development for cybersecurity, researchers from ETH Zürich have identified a critical security flaw affecting all modern Intel processors. This vulnerability, dubbed Branch Privilege Injection (BPI), can lead to unauthorized access to sensitive data stored in memory, amplifying concerns around the long-standing Spectre vulnerabilities that have troubled computer systems for over seven years.

BPI exploits the predictive calculations performed by the Central Processing Unit (CPU), allowing malicious actors to manipulate these processes to extract data from the memory spaces of other users sharing the same processor. Kaveh Razavi, the head of the Computer Security Group (COMSEC) at ETH Zürich and a co-author of the study, emphasized the vulnerability’s broad implications, as it potentially permits attackers to access cached data and operational memory belonging to different users on the affected CPU.

The attack hinges on exploiting Branch Predictor Race Conditions (BPRC), which occur during the CPU’s transition between predictive calculations. This presents a unique avenue for attackers to bypass traditional security measures and gain ephemeral access to confidential information. The ramifications of this flaw could extend far beyond theoretical vulnerabilities, posing a tangible risk to businesses relying on Intel processors in their infrastructure.

As companies increasingly integrate Intel hardware into their systems, the discovery underscores the necessity for vigilant cybersecurity practices. The ongoing issues related to Spectre vulnerabilities demonstrate that attackers continue to develop sophisticated methods to exploit fundamental architecture choices in modern CPUs.

This vulnerability primarily targets any business utilizing affected Intel processors, which are widespread across various industries, making the impact potentially far-reaching. In terms of cybersecurity frameworks, the tactics that could align with this attack scenario may include initial access through user-side vulnerabilities and privilege escalation mechanisms, revealing the layered complexity of modern cyber threats.

While the scale of potential exploitation remains to be fully understood, the emergence of BPI warrants immediate attention from business owners and IT departments. Organizations are urged to evaluate their cybersecurity posture and investigate potential mitigations against such memory leak vulnerabilities.

As the landscape of cyber threats evolves, this incident serves as a crucial reminder of the persistent risks posed by hardware vulnerabilities and the importance of robust security measures in safeguarding sensitive information.

Source link

Related Posts

Russia-Linked APT28 Exploits MDaemon Zero-Day to Target Government Webmail Servers

May 15, 2025
Vulnerability / Email Security

A cyber espionage operation associated with a Russian threat actor is reportedly compromising webmail servers, including Roundcube, Horde, MDaemon, and Zimbra, by exploiting cross-site scripting (XSS) vulnerabilities, notably a zero-day flaw in MDaemon. This activity, coded as Operation RoundPress by ESET, began in 2023 and has been linked with moderate confidence to the state-sponsored hacking group APT28, also known by various names such as BlueDelta, Fancy Bear, and Sednit.

“The primary objective of this operation is to extract sensitive data from targeted email accounts,” stated ESET researcher Matthieu Faou in a report shared with The Hacker News. “While most victims are governmental and defense entities in Eastern Europe, we have also noted targets across Africa, Europe, and beyond.”

Security Flaw in AWS Default IAM Roles Threatens Lateral Movement and Cross-Service Exploitation

Researchers in cybersecurity have identified concerning default identity and access management (IAM) roles within Amazon Web Services (AWS) that could potentially allow attackers to escalate privileges, manipulate other AWS services, and even compromise accounts entirely. According to Aqua researchers Yakir Kadkoda and Ofek Itach, “These roles, typically created automatically or suggested during setup, grant excessively broad permissions, including full access to S3.” They warn that these default roles create silent attack vectors for privilege escalation and cross-service access, leading to possible account breaches. The cloud security firm pinpointed vulnerabilities in default IAM roles established by AWS services such as SageMaker, Glue, EMR, and Lightsail. A similar issue has also been detected in the widely-used open-source framework Ray, which generates a default IAM role (ray-autoscaler-v1) that includes the AmazonS3FullAccess policy.

Russian Hackers Target Ukraine Aid Logistics Through Email and VPN Vulnerabilities

May 21, 2025
Cyber Espionage / Vulnerability

State-sponsored Russian cyber actors have been linked to a campaign focused on Western logistics and tech firms since 2022. This activity is attributed to APT28 (also known as BlueDelta, Fancy Bear, or Forest Blizzard), connected to the Russian GRU’s 85th Main Special Service Center, Military Unit 26165. Key targets include companies involved in the coordination and delivery of international aid to Ukraine, as highlighted in a joint advisory from agencies across Australia, Canada, Czechia, Denmark, Estonia, France, Germany, the Netherlands, Poland, the United Kingdom, and the United States. The bulletin notes that this cyber-espionage campaign employs a range of previously identified tactics and is likely linked to broader efforts aimed at IP cameras in Ukraine and neighboring NATO countries.

Critical Flaws in Versa Concerto Allow Attackers to Escape Docker and Compromise Hosts

May 22, 2025
Vulnerability / Software Security

Cybersecurity researchers have identified several severe vulnerabilities within the Versa Concerto network security and SD-WAN orchestration platform. Exploitation of these flaws could potentially grant attackers control over vulnerable instances. Despite responsible disclosure on February 13, 2025, these issues remain unpatched, leading to a public announcement after the 90-day window expired. According to ProjectDiscovery researchers Harsh Jaiswal, Rahul Maini, and Parth Malhotra, “When combined, these vulnerabilities could enable an attacker to fully compromise both the application and the host system.” The vulnerabilities include:

  • CVE-2025-34025 (CVSS score: 8.6): A privilege escalation and Docker container escape vulnerability resulting from unsafe default mounting of host binary paths, potentially allowing code execution on the host system.