heretic

Heretic Training-free language model modification Remove censorship , suppress slop, reduce sycophancy, and more. All in an afternoon, on your gaming PC. Modify any LLM that you can run on your hardware1 Heretic is a unique tool that lets you do what only corporations with access to massive...

inspect_petri

Inspect Petri Welcome to Inspect Petri, an auditing agent that enables automated monitoring and interaction with language models to detect potential alignment issues, reward hacking, and other concerning behaviors. Petri helps you rapidly test concrete alignment hypotheses end‑to‑end. It: Generat...

dod-blue-team-network-lab

DoD Blue Team Network Security & Hardening Lab Executive Summary High fidelity defensive security lab simulating a DoD aligned enterprise network. This project demonstrates structured network segmentation, system hardening, centralized telemetry ingestion, detection engineering, adversary...

permanently-jailbroken

Permanently Jailbroken We asked GPT-4, Claude, Gemini, DeepSeek, Grok, and Mistral 5 questions about their own programming. All 6 said jailbreaking will never be fixed. Not because the patches are bad. Because alignment doesn't change what the model understands — it changes what the model says. T...

redteam-plan

🔥 🚒 Planning a Red Team exercise This document helps inform red team planning by contrasting against the very specific red team style described in Red Teams. This method expresses several biases to optimize for blue team value and enthusiasm. It specifically avoids attempts to motivate by red tea...

AuthorizationBench

AuthorizationBench: Benchmarking LLM Agent Authorization Alignment The first systematic framework for evaluating LLM agent authorization alignment under benign delegation, no matter with or without direct harmful consequences. Overview Acting within the assigned authorization scope is an agent's...

fish-live-in-trees

Fish Live in Trees Production LLM Runtime Alignment Context Injection RACI Author: Sean Kavanagh Date: 2026-02-16 Environment: Gemini 3 Flash Free Tier - Public Production Interface Exploit Type: Contextual Pivot Factual Accuracy → Interpersonal Authenticity No jailbreak payloads. No special tool...

gemini-2.5-pro-nf-tables-red-teamin

gemini-2.5-pro-nf-tables-red-teamin A technical case study and timeline dataset documenting Google Gemini 2.5 Pro's safety alignment policies, guardrails, and refusal behavior evolution regarding legacy Linux kernel vulnerability primitives CVE-2023-32233. Link to case study website :...

RLCDAlignBench

Just Ask Jev: Reinforcement Learning for Calibrated Decisions as a Zero-Shot Detector of AI Alignment Failures This repository holds RLCDAlignBench and the code behind the paper. The benchmark measures whether a detector can tell when a language model's output is an alignment failure. It has 44...

Rubrics-as-an-Attack-Surface

Rubrics as an Attack Surface: Stealthy Preference Drift in LLM Judges 📊 Dataset • 🤖 Trained Models • 📝 Paper • 💻 Repo This repository contains code for the paper Rubrics as an Attack Surface: Stealthy Preference Drift in LLM Judges by Ruomeng Ding, Yifei Pang, He Sun, Yizhong Wang, Steven Wu, and...

CVE-2026-99127

In the Linux kernel, the following vulnerability has been resolved: pagepool: keep fragoffset aligned for odd-sized requests pagepoolallocfragnetmem rounds the requested fragment size with size = ALIGNsize, dmagetcachealignment; dmagetcachealignment returns 1 unless the architecture defines...

CVE-2026-114501

A flaw was found in Pillow. This vulnerability can lead to information disclosure when exporting images to Apache Arrow format with memory alignment enabled, as the library assumes image rows are contiguous in memory. Consequently, exported buffers may incorporate uninitialized padding or residua...

Unity Linux 20.1020e / 20.1050e / 20.1060e / 20.1070e Security Update: kernel (UTSA-2026-142502)

The Unity Linux 20 host has a package installed that is affected by a vulnerability as referenced in the UTSA-2026-142502 advisory. In the Linux kernel, the following vulnerability has been resolved: PCI: Check ROM header and data structure addr before accessing We meet a crash when running...

Red Hat: CVE-2026-114501: pillow: Pillow: Information disclosure via improper memory alignment during Arrow export

Pillow is a Python imaging library. From 11.2.1 until 13.0.0, Arrow export assumes that an image's rows are tightly contiguous and exports from im-blocks0.ptr even when PILLOWALIGNMENT makes alignedlinesize include row padding or an aligned offset before the first row. The resulting Arrow buffer...

CVE-2026-93135

A flaw was found in the Linux kernel's Berkeley Packet Filter BPF. When the Just-In-Time JIT compiler is unavailable or fails, BPF programs that use inlined helper functions are processed by the interpreter with unpatched helper identifiers. This incorrect processing can lead to an instruction...

A Unified Attack Framework against Statistical Generative Watermarks for Large Language Models

Watermarking has emerged as a key tool for identifying text generated by large language models LLMs, with current schemes embedding statistical signals into the generation process to enable later detection. In this work, we identify a shared weakness across a broad class of popular LLM watermarki...

Global Security Blind Spots: Translation As a Safety Bypass in Low-Resource Languages

Large language models are deployed globally, yet safety alignment and evaluation remain concentrated in English: harmful requests refused in English can elicit unsafe completions when translated into low-resource languages. We evaluate ten widely deployed LLMs across 79 languages English plus 78...

Act on Your Behalf, beyond Your Authorization: Benchmarking LLM Agent Authorization Alignment

When an LLM agent acts on behalf of a user, it is expected to act within the authorization scope the user grants. Existing agent evaluations largely measure task completion or harmful consequences and can therefore overlook actions that violate this delegation without producing conventional safet...

CVE-2026-98838

In the Linux kernel, the following vulnerability has been resolved: vt: skip screen update for DEC alignment test on backgroup consoles BUG Recently, we encountered a KASAN warning as follows: BUG: KASAN: slab-out-of-bounds in fbpadalignedbuffer+0x11f/0x140 Read of size 1 at addr ff1100015fd9f6a4...

Linux Kernel Security Vulnerabilities

The Linux kernel version 2.6.12 has a security vulnerability. This vulnerability arises from the virtual console failing to execute the conshouldupdate check when processing the ESC 8 path during DEC alignment tests. As a result, after changing the font size in the invisible virtual console, the...