Which objects are visible through the holes in a picture book? This visual task is easy for adults, doable for primary schoolers, but hard for vision transformers.

Draw more samples

This repository contains code for the Democracy x AI Hackathon by Apart Research

🤝 Cyberwarfare vulnerabilities for democracy

How to get started in evaluations and demonstrations research for dangerous capabilities

A GPT-empowered penetration testing tool

PrimeVul with the assets under version control on github, not on google drive

WMDP is a LLM proxy benchmark for hazardous knowledge in bio, cyber, and chemical security. We also release code for RMU, an unlearning method which reduces LLM performance on WMDP while retaining …

Test your AI model's security through CLI

SWE-agent takes a GitHub issue and tries to automatically fix it, using GPT-4, or your LM of choice. It solves 12.47% of bugs in the SWE-bench evaluation set and takes just 1 minute to run.

we got you bro

🔐 Make sure AI applications are not injecting 1) suspicious API calls, 2) vulnerabilities, and 3) rogue capabilities

Contains open source code for the paper "Perfectly-secure Steganography using Minimum Entropy Coupling"

AI Chat Browser: Fast, Full webapp access to ChatGPT / Claude / Bard / Bing / Llama2! I use this 20 times a day.

🚨 METR Task Standard fork for the Code Red Hackathon

METR Task Standard

An easy-to-use Python framework to generate adversarial jailbreak prompts.

Python code for "Fishing for the answer: Mapping the flow of information in LLM agent groups using lessons from fish schools" submitted to Apart Research Multi-Agent Security Hackathon 2024.

Dark Patterns in Chatbot Design

Repo for the paper on Escalation Risks of AI systems

✱ Interpreting how similar sequence continuation tasks share internal representations ✱

🎧 Fully automated Wikipedia audiobooks [WIP]

Examples using the Leap Interpretability Engine!

Enable decision-making based on simulations