Taxonomy of Failure Mode in Agentic AI Systems
A Microsoft AI Red Team taxonomy of safety and security failure modes in agentic AI systems, combining a failure mode and effects analysis with mitigation guidance and a memory poisoning case study.
AI Governance Library: all resources
Every reviewed AI governance resource in one place — reports, standards, guidance and regulation.
State of AI Policy in Africa 2025
A twelve-point rubric assessment of AI policy maturity across twenty African countries as of October 2025, scoring policy design, implementation capacity, governance and accountability, and external engagement out of 24.
Frontier Safety Framework Version 2.0
Google DeepMind's Frontier Safety Framework version 2.0 sets out protocols for detecting Critical Capability Levels in frontier models and applying security and deployment mitigations, with tables pairing misuse and deceptive alignment thresholds to mitigations.
Frontier Safety Framework
Google's Frontier Safety Framework Version 3.1 sets out Critical Capability Levels and Tracked Capability Levels for misuse (CBRN, cyber, harmful manipulation) and ML R&D and misalignment risks, with security and deployment mitigations, a risk management process and governance.
Microsoft Responsible AI Standard, v2: General Requirements for External Release
Microsoft's second version of its Responsible AI Standard, setting out the general requirements its AI systems must meet across accountability, transparency, fairness, reliability and safety, privacy, security and inclusiveness goals.
AI Assurance: A Repeatable Process for Assuring AI-enabled Systems
MITRE executive summary of a technical paper setting out a four-step AI assurance process—discovering assurance needs, characterizing and prioritizing risks, evaluating risks, and managing risks—that produces an AI Assurance Plan.
AI Assurance – Challenges, Maturity, and Paths Forward
MITRE policy wrapper summarising its 6 June 2024 AI Assurance Summit, where more than 150 federal attendees discussed AI assurance challenges, technical capability maturity and paths forward including assurance labs and automation.
A Sensible Regulatory Framework for AI Security
MITRE paper proposing elements of a US regulatory framework for AI security, organised around three categories of AI application — subsystem, human augmentation and agency — with twelve recommendations on assurance, auditability, transparency and critical infrastructure.
The AI Assurance Landscape (v1.0): Toward a Standardized Framework of Unique and Differentiated AIA Concepts
MITRE report (v1.0) presenting the AI Assurance Landscape, a concept map of 11 AIA categories and 66 specific assurance needs synthesised from roughly 50 existing AI assurance frameworks and reports, with an accompanying glossary of definitions and cross-framework mappings.
Assuring AI Security and Safety Through AI Regulation
MITRE Center for Data-Driven Policy presidential transition memo proposing nine data-driven recommendations for a US regulatory framework covering AI assurance, auditability, transparency, incident sharing and governance.
Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems
An arXiv preprint from the AI Standards Lab presenting a catalog of risk sources and risk management measures for general-purpose AI systems across development, evaluation, deployment, cybersecurity and impact stages.
Risk Discovery Protocol for AI Assurance (v1.0): Guidance for Administering the Protocol
MITRE guidance for administering the Risk Discovery Protocol for AI Assurance (RDP-AIA), a four-phase interview-based methodology that helps teams unpack a mission problem and AI solution, elicit AI assurance needs, and prioritize them by assessed risk in a Priority Report.
Strategic review of the whole of Australian Government single seller arrangements
Independent strategic review prepared by SolsticeIT for the Digital Transformation Agency, assessing whether the Australian Government's Single Seller Arrangements with AWS, IBM, Microsoft, Oracle, Rimini Street and SAP remain fit-for-purpose and deliver value.
Policy for the responsible use of AI in government
Australian Government policy setting out how agencies adopt and use AI, with an 'enable, engage and evolve' framework, mandatory requirements on accountable officials, high-risk use case notification and AI transparency statements, plus a risk matrix.
Policy for the responsible use of AI in government
Australian Government policy (version 2.0) setting principles and mandatory requirements for agencies' use of AI, covering strategy and oversight, preparedness and operations, and AI use case impact assessment.
Responsible AI in the automotive industry: Techniques and use cases
A joint Accenture and DFKI paper on Responsible AI in the automotive industry. It sets out Accenture's seven Responsible AI principles, links them to AI maturity, introduces neuro-explicit AI and works through seven autonomous driving use cases.
National framework for the assurance of artificial intelligence in government: A joint approach to safe and responsible AI by the Australian, state and territory governments
A framework from the Australian, state and territory governments setting out a nationally consistent approach to assuring government use of AI. It defines cornerstones of assurance and maps practices to Australia's eight AI Ethics Principles.
Explanatory Report to the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law
Article-by-article commentary on the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law (CETS 225), published by the Council of Europe.
Guidance for AI adoption: implementation guidance
Australian Government implementation guidance setting out six responsible AI adoption practices for developers and deployers, covering accountability, risk management, transparency, testing and human control, with definitions and a VAISS crosswalk.
Governing with Artificial Intelligence: The State of Play and Way Forward in Core Government Functions
An OECD report analysing how governments develop and use AI across core public functions, based on 200 use cases and dozens of governance approaches. It sets out opportunity areas, five risk types specific to government, and a framework of enablers, guardrails and engagement.
Red Teaming Artificial Intelligence for Social Good: The PLAYBOOK
A UNESCO playbook, written with Humane Intelligence, that guides organizations and communities through designing and running red teaming exercises on generative AI models to expose bias, stereotypes and technology-facilitated gender-based violence.
Digital Government Outlook 2026: From Foundations to Transformational Impact
First OECD Digital Government Outlook, assessing digital government across 36 OECD members and 8 accession candidate countries through the 2025 Digital Government Index and OURdata Index. Covers digital public infrastructure, data governance, investment, skills, AI adoption and service delivery.
AI Governance Alliance Briefing Paper Series
The AI Governance Alliance Briefing Paper Series bundles three January 2024 briefing papers on safe generative AI models, responsible enterprise transformation, and international AI governance, interoperability and equitable access.
Unlocking Value from Generative AI: Guidance for Responsible Transformation
World Economic Forum briefing paper, produced with IBM Consulting, on how organizations can capture value from generative AI use cases while managing downstream impacts. It sets out three evaluation gates for selecting use cases and four functions for responsible transformation.
