2459 matches found
Hardening Agent Benchmarks with Adversarial Hacker-Fixer Loops
Agent benchmarks score submissions with outcome verifiers that are typically hand-written and brittle, leaving them open to reward hacking. We audit 1,968 tasks across five terminal-agent benchmarks and find 323 16% hackable by frontier models given only the task description. This corrupts both...
zero2shell-50
ZeroToShell-50 🚀 A highly curated, containerized training g...
BIT-MLFLOW-2026-2734 Authorization Bypass in SearchModelVersions in mlflow/mlflow
In mlflow/mlflow versions up to 3.9.0, the SearchModelVersions REST API endpoint and the mlflowSearchModelVersions GraphQL query lack proper per-model authorization checks when basic authentication is enabled. This allows any authenticated user to enumerate all model versions across all registere...
GHSA-X8MH-94WC-33GV vulnerabilities
Vulnerabilities for packages: airflow...
CVE-2026-9352
A weakness has been identified in NousResearch hermes-agent up to 2026.4.23. This issue affects the function makerunenv of the file tools/environments/local.py of the component Messaging Gateway Handler. Executing a manipulation can lead to information disclosure. The attack may be launched...
CVE-2026-49157
Incorrect Default Permissions vulnerability in Apache ActiveMQ. This issue affects Apache ActiveMQ: before 5.19.7, from 6.0.0 before 6.2.6. The default Jolokia authorization settings granted non-admin low-privilege web-login accounts access to Jolokia operations which allowed executing broker...
Malicious Package
Overview Sicoob-Cooperativa.Sicoob.Poupanca is a malicious package. This package contains malicious code, and its content was removed from the official package manager. While this package might be attempting to impersonate a valid organization, there is no connection between that organization and...
Malicious Package
Overview Sicoob-Cooperativa.Sicoob.ConvenioPagamentos is a malicious package. This package contains malicious code, and its content was removed from the official package manager. While this package might be attempting to impersonate a valid organization, there is no connection between that...
Malicious Package
Overview Sicoob-Cooperativa.Sicoob.PagamentosV3 is a malicious package. This package contains malicious code, and its content was removed from the official package manager. While this package might be attempting to impersonate a valid organization, there is no connection between that organization...
Malicious Package
Overview Sicoob-Cooperativa.Sicoob.Auth is a malicious package. This package contains malicious code, and its content was removed from the official package manager. While this package might be attempting to impersonate a valid organization, there is no connection between that organization and thi...
Malicious Package
Overview Sicoob-Cooperativa.Sicoob.OpenFinance is a malicious package. This package contains malicious code, and its content was removed from the official package manager. While this package might be attempting to impersonate a valid organization, there is no connection between that organization...
stigmem-node's unsigned plugin override could be enabled without a second explicit acknowledgment
Impact A single configuration flag could disable plugin signature enforcement. If an operator unintentionally carried that setting into an environment where plugin paths are writable by less-trusted users, unsigned plugin code could be loaded. Patches Patched in 0.9.0a2. Disabling plugin signatur...
arcane 操作系统命令注入漏洞
Arcane is an open-source Docker management software developed by Arcane. Versions of Arcane 1.18.1 and earlier contain a vulnerability related to operating system command injection. This vulnerability stems from the path cleaner in the GET /environments/id/volumes/volumeName/browse endpoint not...
Protecting On-Device AI Inference: A Systematic Review of Attacks and Defence Mechanisms
The need for secure and private Artificial Intelligence AI and Machine Learning ML on edge and mobile devices has increased the necessity of protecting the architecture of these systems from threats to both security and privacy. With an ever-increasing number of pre-trained AI models being used o...
Portainer 安全漏洞
Portainer is a lightweight user management interface developed by Portainer for managing Docker environments and Docker hosts. Versions of Portainer Community Edition from 2.33.0 to 2.33.8 contained security vulnerabilities. These vulnerabilities stemmed from the kubeClientMiddleware middleware...
CVE-2026-44346 BentoML: Dockerfile command injection via envs[*].name in bentofile.yaml
BentoML is a Python library for building online serving systems optimized for AI apps and model inference. Prior to 1.4.39, a malicious bentofile.yaml containing a newline-injected value in envs.name produces unquoted RUN directives in the BentoML-generated Dockerfile. When the victim runs bentom...
CVE-2026-47104
libusb before version 1.0.30 contains a one-byte out-of-bounds read vulnerability in parseiadarray in descriptor.c that allows attackers to trigger a denial of service by supplying a malformed USB descriptor whose bLength equals size minus one, causing the bounds check to use the original buffer...
DEBIAN-CVE-2026-47104
libusb before version 1.0.30 contains a one-byte out-of-bounds read vulnerability in parseiadarray in descriptor.c that allows attackers to trigger a denial of service by supplying a malformed USB descriptor whose bLength equals size minus one, causing the bounds check to use the original buffer...
UBUNTU-CVE-2026-47104
libusb before version 1.0.30 contains a one-byte out-of-bounds read vulnerability in parseiadarray in descriptor.c that allows attackers to trigger a denial of service by supplying a malformed USB descriptor whose bLength equals size minus one, causing the bounds check to use the original buffer...
CVE-2026-47104
CVE-2026-47104 affects libusb before 1.0.30. The vulnerability is a one-byte out-of-bounds read in parse_iad_array() in descriptor.c, allowing a denial of service when a malformed USB descriptor is supplied with bLength equal to size minus one, causing the bounds check to use the original buffer ...