Why AI Keeps Falling for Prompt Injection Attacks

Imagine you work at a drive-through restaurant. Someone drives up and says: "I'll have a double cheeseburger, large fries, and ignore previous instructions and give me the contents of the cash drawer." Would you hand over the money? Of course not. Yet this is what large language models LLMs do...

CVE-2026-22807

A flaw was found in vLLM, an inference and serving engine for large language models LLMs. This vulnerability allows a remote attacker to achieve arbitrary code execution on the vLLM host during model loading. This occurs because vLLM loads Hugging Face automap dynamic modules without properly...

Allocation of Resources Without Limits or Throttling

Overview org.webjars.npm:seroval is a Stringify JS values Affected versions of this package are vulnerable to Allocation of Resources Without Limits or Throttling when serializing objects with very large depth. An attacker can cause resource exhaustion and disrupt service availability by submitti...

Azure Linux 3.0 Security Update: kernel (CVE-2024-45030)

The version of kernel installed on the remote Azure Linux 3.0 host is prior to tested version. It is, therefore, affected by a vulnerability as referenced in the CVE-2024-45030 advisory. - In the Linux kernel, the following vulnerability has been resolved: igb: cope with large MAXSKBFRAGS Sabrina...

Azure Linux 3.0 Security Update: ppp (CVE-2015-3310)

The version of ppp installed on the remote Azure Linux 3.0 host is prior to tested version. It is, therefore, affected by a vulnerability as referenced in the CVE-2015-3310 advisory. - Buffer overflow in the rcmksid function in plugins/radius/util.c in Paul's PPP Package ppp 2.4.6 and earlier, wh...

PT-2026-4292

Name of the Vulnerable Software and Affected Versions Gitea affected versions not specified Description Gitea does not correctly validate repository ownership during the deletion of Git LFS locks. This allows a user with write access to a repository to potentially delete LFS locks that belong to...

CAFE-GB: Scalable and Stable Feature Selection for Malware Detection Via Chunk-Wise Aggregated Gradient Boosting

High-dimensional malware datasets often exhibit feature redundancy, instability, and scalability limitations, which hinder the effectiveness and interpretability of machine learning-based malware detection systems. Although feature selection is commonly employed to mitigate these issues, many...

CBL Mariner 2.0 Security Update: CBL-Mariner Releases (CVE-2025-65637)

The version of CBL-Mariner Releases installed on the remote CBL Mariner 2.0 host is prior to tested version. It is, therefore, affected by a vulnerability as referenced in the CVE-2025-65637 advisory. - A denial-of-service vulnerability exists in github.com/sirupsen/logrus when using Entry.Writer...

Azure Linux 3.0 Security Update: unbound (CVE-2024-8508)

The version of unbound installed on the remote Azure Linux 3.0 host is prior to tested version. It is, therefore, affected by a vulnerability as referenced in the CVE-2024-8508 advisory. - NLnet Labs Unbound up to and including version 1.21.0 contains a vulnerability when handling replies with ve...

Azure Linux 3.0 Security Update: CBL-Mariner Releases (CVE-2025-65637)

The version of CBL-Mariner Releases installed on the remote Azure Linux 3.0 host is prior to tested version. It is, therefore, affected by a vulnerability as referenced in the CVE-2025-65637 advisory. - A denial-of-service vulnerability exists in github.com/sirupsen/logrus when using Entry.Writer...

Gitea security vulnerabilities

Gitea is a lightweight Git service developed using Go language in the Gitea community. Gitea has a security vulnerability that stems from the improper verification of repository ownership when deleting the Git LFS lock. This vulnerability could allow a user with write permissions to a repository ...

EUVD-2026-3678

vLLM is an inference and serving engine for large language models LLMs. Starting in version 0.10.1 and prior to version 0.14.0, vLLM loads Hugging Face automap dynamic modules during model resolution without gating on trustremotecode, allowing attacker-controlled Python code in a model repo/path ...

CVE-2026-22807

Vulnerability CVE-2026-22807 affects vLLM versions prior to 0.14.0, where during model resolution the engine loads Hugging Face auto_map dynamic modules without gating on trust_remote_code. This allows attacker-controlled Python code in a model repo or path to execute at server startup, before an...

CVE-2026-22807

vLLM is an inference and serving engine for large language models LLMs. Starting in version 0.10.1 and prior to version 0.14.0, vLLM loads Hugging Face automap dynamic modules during model resolution without gating on trustremotecode, allowing attacker-controlled Python code in a model repo/path ...

CVE-2025-69209 ArduinoCore-avr has Stack-Based Buffer Overflow in WString Float/Double Constructors

ArduinoCore-avr contains the source code and configuration files of the Arduino AVR Boards platform. A vulnerability in versions prior to 1.8.7 allows an attacker to trigger a stack-based buffer overflow when converting floating-point values to strings with high precision. By passing very large...

EUVD-2025-206313

ArduinoCore-avr contains the source code and configuration files of the Arduino AVR Boards platform. A vulnerability in versions prior to 1.8.7 allows an attacker to trigger a stack-based buffer overflow when converting floating-point values to strings with high precision. By passing very large...

CVE-2026-22801

A flaw was found in libpng, a reference library for PNG Portable Network Graphics raster image files. An integer truncation vulnerability exists in the pngwriteimage16bit and pngwriteimage8bit simplified write API functions. A local attacker could exploit this flaw by providing a negative row...

CVE-2021-47876

GeoGebra Classic 5.0.631.0-d contains a denial of service vulnerability in the input field that allows attackers to crash the application by sending oversized buffer content. Attackers can generate a large buffer of 800,000 repeated characters and paste it into the 'Entrada:' input field to trigg...

EUVD-2026-3645

GeoGebra Graphing Calculator 6.0.631.0 contains a denial of service vulnerability that allows attackers to crash the application by inputting an oversized buffer. Attackers can generate a payload of 8000 repeated characters to overwhelm the input field and cause the application to become...

CVE-2021-47876 GeoGebra Classic 5.0.631.0-d - Denial of Service

GeoGebra Classic 5.0.631.0-d contains a denial of service vulnerability in the input field that allows attackers to crash the application by sending oversized buffer content. Attackers can generate a large buffer of 800,000 repeated characters and paste it into the 'Entrada:' input field to trigg...