Jailbreak-Tuning: Models Efficiently Learn Jailbreak Susceptibility

AI systems are rapidly advancing in capability, and frontier model developers broadly acknowledge the need for safeguards against serious misuse. However, this paper demonstrates that fine-tuning, whether via open weights or closed fine-tuning APIs, can produce helpful-only models. In contrast to...

FORTRESS: Frontier Risk Evaluation for National Security and Public Safety

The rapid advancement of large language models LLMs introduces dual-use capabilities that could both threaten and bolster national security and public safety NSPS. Models implement safeguards to protect against potential misuse relevant to NSPS and allow for benign users to receive helpful...

Data Breach Reporting for regulatory requirements with Microsoft Data Security Investigations​​

Seventy-four percent of organizations surveyed experienced at least one data security incident with their business data exposed in the previous year as reported in Microsoft’s Data Security Index: Trends, insights, and strategies to secure data report. Despite the best people, process and...

PT-2025-25311

Name of the Vulnerable Software and Affected Versions The product name cannot be determined. Description The issue is related to insufficient measures to prevent multiple failed authentication attempts within a short time frame, making the product susceptible to brute-force attacks. Recommendatio...

Docker under Siege: Securing Containers in the Modern Era

Containerization, driven by Docker, has transformed application development and deployment by enhancing efficiency and scalability. However, the rapid adoption of container technologies introduces significant security challenges that require careful management. This paper investigates key areas o...

The Hidden Dangers of Browsing AI Agents

Autonomous browsing agents powered by large language models LLMs are increasingly used to automate web-based tasks. However, their reliance on dynamic content, tool execution, and user-provided data exposes them to a broad attack surface. This paper presents a comprehensive security evaluation of...

The vulnerability of the ColdFusion software platform, related to the lack of measures to neutralize special elements, allows attackers to execute arbitrary code.

The vulnerability of the ColdFusion software platform is related to the lack of measures taken to neutralize special elements. Exploiting this vulnerability allows a malicious actor to execute arbitrary code remotely...

Unsafe `Deserialization` in `JsonPickleSerializer` Enables Remote Code Execution

Description A critical deserialization vulnerability exists in the llamaindex library’s JsonPickleSerializer component, enabling remote code execution RCE due to an insecure fallback to Python’s pickle module. When deserializing untrusted data, JsonPickleSerializer prioritizes pickle.loads, which...

CVE-2025-20226

In Splunk Enterprise versions below 9.4.1, 9.3.3, 9.2.5, and 9.1.8 and Splunk Cloud Platform versions below 9.3.2408.107, 9.2.2406.111, and 9.1.2308.214, a low-privileged user that does not hold the "admin" or "power" Splunk roles could run a saved search with a risky command using the permission...

CVE-2025-20232 Risky Command Safeguards Bypass in “/app/search/search“ endpoint through “s“ parameter in Splunk Enterprise

In Splunk Enterprise versions below 9.3.3, 9.2.5, and 9.1.8 and Splunk Cloud Platform versions below 9.3.2408.103, 9.2.2406.108, 9.2.2403.113, 9.1.2312.208 and 9.1.2308.212, a low-privileged user that does not hold the “admin“ or “power“ Splunk roles could run a saved search with a risky command...

CVE-2025-20226 Risky command safeguards bypass in “/services/streams/search“ endpoint through “q“ parameter in Splunk Enterprise

In Splunk Enterprise versions below 9.4.1, 9.3.3, 9.2.5, and 9.1.8 and Splunk Cloud Platform versions below 9.3.2408.107, 9.2.2406.111, and 9.1.2308.214, a low-privileged user that does not hold the "admin" or "power" Splunk roles could run a saved search with a risky command using the permission...

CVE-2025-20226 Risky command safeguards bypass in “/services/streams/search“ endpoint through “q“ parameter in Splunk Enterprise

In Splunk Enterprise versions below 9.4.1, 9.3.3, 9.2.5, and 9.1.8 and Splunk Cloud Platform versions below 9.3.2408.107, 9.2.2406.111, and 9.1.2308.214, a low-privileged user that does not hold the "admin" or "power" Splunk roles could run a saved search with a risky command using the permission...

Splunk Enterprise 9.1.0 < 9.1.8, 9.2.0 < 9.2.5, 9.3.0 < 9.3.3, 9.4.0 < 9.4.1 (SVD-2025-0305)

The version of Splunk installed on the remote host is prior to tested version. It is, therefore, affected by a vulnerability as referenced in the SVD-2025-0305 advisory. - In Splunk Enterprise versions below 9.4.1, 9.3.3, 9.2.5, and 9.1.8 and Splunk Cloud Platform versions below 9.3.2408.107,...

Jailbreaking is (mostly) simpler than you think

Content warning: This blog post contains discussions of sensitive topics. These subjects may be distressing or triggering for some readers. Reader discretion is advised. Today, we are sharing insights on a simple, optimization-free jailbreak method called Context Compliance Attack CCA, that has...

A Signal Update Fends Off a Phishing Technique Used in Russian Espionage

Google warns that hackers tied to Russia are tricking Ukrainian soldiers with fake QR codes for Signal group invites that let spies steal their messages. Signal has pushed out new safeguards...

CVE-2024-53244 Risky command safeguards bypass in “/en-US/app/search/report“ endpoint through “s“ parameter

In Splunk Enterprise versions below 9.3.2, 9.2.4, and 9.1.7 and Splunk Cloud Platform versions below 9.2.2406.107, 9.2.2403.109, and 9.1.2312.206, a low-privileged user that does not hold the “admin“ or “power“ Splunk roles could run a saved search with a risky command using the permissions of a...

CVE-2024-53244

CVE-2024-53244 affects Splunk Enterprise and Splunk Cloud Platform: versions prior to 9.3.2 (Enterprise) and prior to 9.2.2406.107, 9.2.2403.109, 9.1.2312.206 (Cloud) are vulnerable. The issue is a bypass of SPL safeguards for risky commands via the s parameter on the /en-US/app/search/report end...

PT-2024-35689 · Splunk · Splunk Cloud Platform +1

Name of the Vulnerable Software and Affected Versions: Splunk Enterprise versions prior to 9.3.2 Splunk Enterprise versions prior to 9.2.4 Splunk Enterprise versions prior to 9.1.7 Splunk Cloud Platform versions prior to 9.2.2406.107 Splunk Cloud Platform versions prior to 9.2.2403.109 Splunk Clo...

Azure AI Vulnerabilities Allowed Attacks to Bypass Moderation Safeguards

Mindgard researchers uncovered critical vulnerabilities in Microsoft's Azure AI Content Safety service, allowing attackers to bypass its safeguards…...

Subnet Solutions Inc. PowerSYSTEM Center

View CSAF 1. EXECUTIVE SUMMARY CVSS v3 7.5 ATTENTION: Exploitable remotely/Low attack complexity Vendor: Subnet Solutions Inc. Equipment: PowerSYSTEM Center Vulnerabilities: Server-Side Request Forgery SSRF, Inefficient Regular Expression Complexity, Cross-Site Request Forgery CSRF 2. RISK...