A compact, practical guide to understanding the most influential cognitive biases in everyday thinking, decision-making, and AI design—plus a bonus chapter on algorithmic bias. A must-read for anyone working at the intersection of technology, governance, and ethics.
Simon Mylius’s Scalable AI Incident Classification showcases a proof-of-concept system using large language models (LLMs) to summarize, classify, and rate AI incidents at scale.
Amlan Mohanty’s report, Making AI Self-Regulation Work, offers a comprehensive framework for deploying self-regulation as a foundational piece of India’s AI governance.
Open source AI is no longer fringe—it’s becoming essential. Based on a global survey of over 700 tech leaders, this report shows how open source AI is reshaping tech stacks, boosting developer satisfaction, and challenging proprietary dominance.
A highly structured guide showing how to apply OWASP’s Agentic AI Threat Taxonomy to real-world multi-agent systems (MAS), introducing the MAESTRO framework to surface layered vulnerabilities and new attack paths unique to agent-to-agent coordination environments.
This issue is about the foundations no one celebrates—but everyone depends on. Solid spreadsheets, clean templates, tools that won’t break under pressure. Plus: models planning backwards, AI apps built in a weekend, and how a book cover can teach you everything about framing AI risk.
The AI Risk Assessment Template provides a structured, highly practical checklist for evaluating AI system risks across development, deployment, and operation phases. It aligns with NIST AI RMF and EU AI Act requirements, aiming to boost trustworthy AI practices.
This week we look at how visual design can make or break AI governance, review the first real playbook for agent oversight, and highlight why Tom Scott’s YouTube channel might teach you more about systems failure than most compliance workshops ever could.
This guide explores how to govern autonomous AI agents—systems capable of planning and acting with minimal instruction. It presents a structured approach to agent risks and interventions, pushing the conversation beyond foundation models toward emergent systems.
This workbook is a facilitator’s guide to delivering AI ethics training across public institutions. It covers AI fundamentals, public sector use cases, and governance models—paired with activities grounded in UK government experience and policy frameworks.
This guidance from New South Wales outlines role-specific responsibilities for implementing responsible AI. It supports public agencies in assigning accountability using ISO-aligned frameworks and practical RACI structures. A useful anchor for everyday governance.
Governance needs clarity—not clickbait. This issue breaks down what makes a good AI governance resource, why Bird & Bird’s AI Act guide sets the bar, and how fake AI cults, Roman GIS maps, and server-blessing priests say more about tech than most frameworks do.
This UK government strategy outlines a six-point digital reform plan focused on service redesign, AI integration, shared infrastructure, and leadership reform. It introduces a new digital centre of government and pushes for transparency, efficiency, and public trust.
The Bird & Bird guide to the EU AI Act offers a deep dive into the Act’s legal obligations, scope, governance model, and technical standards. It walks readers through implementation timelines, roles across the AI value chain, and penalties for non-compliance.
This report, produced under the EDPB’s Support Pool of Experts (SPE) programme, offers structured guidance on managing privacy risks in LLM systems. It lays out risk identification, evaluation, and control strategies tailored to GDPR and AI Act obligations, supporting both developers and deployers.
This chart-based explainer by the IAPP (April 2025) breaks down incident notification and information-sharing obligations under 11 key EU digital laws, helping organisations navigate complex breach scenarios across privacy, cybersecurity, AI, and operational resilience domains.
This is issue #1 of the AIGL newsletter. In the first 24 hours, 382 of you subscribed. In this issue: • A personal note on why this project exists • A review of MIT’s AI Risk Framework—one of the cleanest tools we’ve seen so far • And two sharp links from inside China’s evolving AI landscape
A visual guide and explanatory article by James Kavanagh, published via The Company Ethos (April 2025), that maps major AI risks—like bias, hallucinations, and adversarial attacks—to practical prevention, detection, and response controls across design-time and run-time phases.
This is one of the best open-source resources for operationalizing multi-standard AI governance. Kavanagh’s framework deserves a place on every AI compliance lead’s desk.
A tool built for clarity, not complexity. The AI Risk Analysis Framework from MIT offers a structured, policy-relevant approach to thinking about AI risks. It’s designed for teams who need to assess potential harms without getting buried in technical noise.
A practical tool to support procurement teams, legal advisors, CISOs, and AI governance leads in evaluating vendors supplying AI-based systems or services.
That’s the question this report tackles—offering one of the most detailed attempts so far to break down “trustworthy AI” into 150 actionable properties across the AI lifecycle. The taxonomy isn’t just a list.
This report breaks down why that question isn’t so easy to answer when AI is built and deployed by many hands. It proposes a framework for assigning civil liability across the AI value chain.
Even when pretraining data is clean, large language models can still absorb—and amplify—political bias. This study maps how that happens and how it affects hate speech and misinformation detection.
Curated Library of AI Governance Resources