vger

V'ger V'ger is an interactive command-line application for post-exploitation of authenticated Jupyter instances with a focus on AI/ML security operations. User Stories As a Red Teamer , you've found Jupyter credentials, but don't know what you can do with them. V'ger is organized in a format that...

GenAI-LLM-Top10

OWASP Top 10 for Large Language Model Applications !IMPORTANT Current release: 2026 — published August 4, 2026. Get the 2026 publication · Read the canonical source · Report a correction This is the official source repository for the OWASP Top 10 for Large Language Model Applications , maintained...

ToBAC

🚬 ToBAC NeurIPS 2026 NeurIPS 2026 · Official PyTorch implementation Token by Token, Compromised: Backdoor Vulnerabilities in Unified Autoregressive Models ToBAC Token by Token Backdoor Attack studies how backdoors affect unified autoregressive models that generate both images and text. We...

StratFL: Countering Covert Model Poisoning Attacks with Strategic Federated Learning

Advanced persistent threats APTs are a major concern in federated learning FL deployments in critical sectors such as healthcare, and finance. Traditional defense mechanisms fail to contain APTs due to their intermittent attacking behavior. This paper addresses a specific type of APT called cover...

Compression Footprints As Security Signals for Model-Poisoning Defense in Federated Learning

Lossy compression is widely used in Federated Learning FL but is generally treated as an error source, while conventional poisoning defenses inspect update geometry. In this work, we instead treat the compressor's response as a security signal: the input-dependent distortion and payload behavior...

A Malicious Webpage Could Poison Your Local AI Model Behind NVIDIA NemoClaw

Oasis Security has disclosed a weakness in NVIDIA NemoClaw that could let an attacker-controlled webpage take unauthenticated control of the local Ollama instance serving an AI agent and plant hidden instructions inside the model itself. The findings were shared with The Hacker News ahead of...

PT-2026-81516

Name of the Vulnerable Software and Affected Versions NVIDIA NemoClaw versions 0.0.0 through 0.0.25 Description NVIDIA NemoClaw for Linux contains a flaw in its inference server setup that allows a remote attacker to access the inference service without authentication. In affected configurations,...

EUVD-2026-62958

In Splunk AI Toolkit versions below 6.0.0, a user who can upload models could overwrite a model being uploaded by another user by sending a concurrent upload request for the same model name, causing the resulting model lookup entry to reference attacker-controlled content. The race condition is...

GHSA-FFJ6-66C4-86GW NLTK downloader allows cross-package resource and model poisoning

A vulnerability in nltk.downloader in nltk/nltk versions = 3.9.4 allows for cross-package resource and model poisoning. The downloader extracts package archives into shared namespaces such as corpora/ and taggers/ instead of package-isolated roots, and validates package integrity only after the...

NLTK downloader allows cross-package resource and model poisoning

A vulnerability in nltk.downloader in nltk/nltk versions = 3.9.4 allows for cross-package resource and model poisoning. The downloader extracts package archives into shared namespaces such as corpora/ and taggers/ instead of package-isolated roots, and validates package integrity only after the...

CVE-2026-12261

A vulnerability in "nltk.downloader" in nltk/nltk versions = 3.9.4 allows for cross-package resource and model poisoning. The downloader extracts package archives into shared namespaces such as "corpora/" and "taggers/" instead of package-isolated roots, and validates package integrity only after...

CVE-2026-12261: Improper Access Control

A vulnerability in nltk.downloader in nltk/nltk versions = 3.9.4 allows for cross-package resource and model poisoning. The downloader extracts package archives into shared namespaces such as corpora/ and taggers/ instead of package-isolated roots, and validates package integrity only after the...

PT-2026-105507

A vulnerability in nltk.downloader in nltk/nltk versions = 3.9.4 allows for cross-package resource and model poisoning. The downloader extracts package archives into shared namespaces such as corpora/ and taggers/ instead of package-isolated roots, and validates package integrity only after the...

PT-2026-68874

Name of the Vulnerable Software and Affected Versions nltk versions prior to 3.9.5 Description A flaw in nltk.downloader allows for cross-package resource and model poisoning. The downloader extracts package archives into shared namespaces, such as corpora/ and taggers/, rather than using...

CVE-2026-12261

A vulnerability in nltk.downloader in nltk/nltk versions = 3.9.4 allows for cross-package resource and model poisoning. The downloader extracts package archives into shared namespaces such as corpora/ and taggers/ instead of package-isolated roots, and validates package integrity only after the...

Evading Chain-Of-Thought Monitoring through Model Poisoning

Chain-of-thought CoT monitoring is an increasingly important component of AI safety stacks but relies on the assumption that a model's reasoning trace is informative about its actions. This work studies the limits of CoT monitoring through the lens of model poisoning. We demonstrate that backdoor...

PT-2026-59163

Predictable bucket naming in Vertex AI Experiments in Google Cloud Vertex AI from version 1.21.0 up to but not including 1.133.0 on Google Cloud Platform allows an unauthenticated remote attacker to achieve cross-tenant remote code execution, model theft, and poisoning via pre-creating predictabl...

Undefined Behavior in mlflow

A vulnerability in mlflow/mlflow version 2.11.1 allows attackers to create multiple models with the same name by exploiting URL encoding. This flaw can lead to Denial of Service DoS as an authenticated user might not be able to use the intended model, as it will open a different model each time...

PYSEC-2026-1645 Undefined Behavior in mlflow

A vulnerability in mlflow/mlflow version 2.11.1 allows attackers to create multiple models with the same name by exploiting URL encoding. This flaw can lead to Denial of Service DoS as an authenticated user might not be able to use the intended model, as it will open a different model each time...

EUVD-2025-210423

Picklescan before 0.0.33 fails to detect the numpy.f2py.crackfortran.getlincoef gadget in pickle reduce methods, allowing arbitrary code execution. Attackers can craft malicious pickle files that execute arbitrary Python code when loaded, bypassing Picklescan's safety checks and enabling...