UK Public Sector Code by Organisation

Inspect: A framework for large language model evaluations

Collection of evals for Inspect AI

ControlArena is a collection of settings, model organisms and protocols - for running control experiments.

Extract residual-stream activations and apply steering vectors (including activation oracles) to any vLLM model during inference.

An Inspect extension for agentic cyber evaluations

A Kubernetes sandbox environment for use with inspect_ai

Reproducing "Natural Emergent Misalignment from Reward Hacking" (MacDiarmid et al., Anthropic 2025) with open-source models. Includes reward-hackable RL environments, misalignment evaluations, training configs, and evaluation scripts. Models trained on OLMo (7B, 32B) and GPT-OSS (20B, 120B).

Accompanying code for Async Control: Stress-testing Asynchronous Control Measures for LLM Agents paper

A Proxmox sandbox environment for use with inspect_ai

An EC2 sandbox environment for use with inspect_ai

Report Official Development Assistance (RODA)

Energy Label Generation Service

A proof-of-concept eval testing whether models continue with (or mention) pre-filled malign behaviours.

This repository created for public search portal

Granting Authority Schema Service

Source code report for METS (Manage your UK Emissions Trading Scheme reporting service)

Application code for the UK Emissions Trading Scheme Registry

Lie Detectors for the Did you lie? paper

Bulk upload of subsidery awards into data base by using excel file

Isambard Containers

Climate Change Agreements service application code

CHPQA Application Repository

A Python library for running AI debate experiments — orchestrate LLM debaters, judges, datasets, and protocols, and measure whether debate improves AI judge accuracy.

Kosmos: An AI Scientist for Autonomous Discovery - An implementation and adaptation to be driven by Claude Code or API - Based on the Kosmos AI Paper - https://arxiv.org/abs/2511.02824

AI identity disclosure evaluations for text and speech interactions