AI Center › AI Safety, Alignment & Security
1ArtificialIntelligence · The Global AI Ecosystem
AI Safety, Alignment & Security
AI Safety, Alignment & Security maps the institutions, frameworks and records that govern serious risks from advanced AI — the national and international safety and security institutes, the frontier developers' own published safety frameworks, the incident and harm databases, and the adversarial-threat taxonomies — each linked to the authority that publishes it.
This is a switchboard: it attributes and points to the primary sources. It does not rank them, rate them, or add an assessment of its own. Every entry names its issuing authority, links to that authority's own page, and records when it was verified.
The bodies and developers listed here publish their own evaluations, frameworks and records. The AI Center gathers them in one neutral, sourced place and points you to each; it does not re-originate, rank or assess their work.
AI Safety & Security Institutes and Intergovernmental Bodies
Government and intergovernmental bodies that evaluate advanced-AI risk and coordinate internationally.
International AI Safety Report 2026
The February 2026 second full edition of the international, expert-authored synthesis of scientific evidence on the capabilities and risks of general-purpose AI, commissioned after the 2023 Bletchley AI Safety Summit and backed by around 30 countries plus the EU, OECD and UN.
AI Security Institute (AISI), United Kingdom
The UK government research organisation, formerly the UK AI Safety Institute, that conducts technical evaluations and research on serious risks from frontier AI.
Center for AI Standards and Innovation (CAISI), United States
The U.S. government center, formerly the U.S. AI Safety Institute, that serves as industry's primary point of contact within NIST for testing, standards and collaborative research on commercial AI systems.
International Network of AI Safety Institutes
An intergovernmental network of national AI safety/security institutes and mandated offices, launched in San Francisco on 21 November 2024, that coordinates technical research and evaluation on advanced AI risks.
Japan AI Safety Institute (Japan AISI)
Japan's national AI Safety Institute, which develops safety evaluation methods, guidance and standards for AI and represents Japan in the international network.
European AI Office
The European Commission body that implements and enforces the EU AI Act, including rules for general-purpose AI models, and represents the EU in the International Network of AI Safety Institutes.
Frontier AI Developer Safety Frameworks & Commitments
The safety frameworks the frontier developers publish themselves, and the government commitment they were published under. Shown side by side; not compared.
Responsible Scaling Policy (RSP), Version 3.4
Anthropic's public policy defining AI Safety Level (ASL) capability thresholds and the security and deployment safeguards it commits to implement before training or releasing models that reach them.
Preparedness Framework, Version 2
OpenAI's document describing how it tracks frontier capabilities (biological/chemical, cybersecurity, AI self-improvement) and the 'High' and 'Critical' capability thresholds that trigger required safeguards before deployment.
Frontier Safety Framework (FSF), Version 3.1
Google DeepMind's framework setting out Critical Capability Levels and Tracked Capability Levels for advanced models, and the evaluation-and-mitigation process it applies as models approach those levels.
Advanced AI Scaling Framework, Version 2 (formerly the Frontier AI Framework)
Meta's framework describing how it identifies catastrophic-risk outcomes (chemical/biological, cybersecurity, and loss of control), classifies models by risk level, and decides whether to release, restrict, or halt development.
Amazon's Frontier Model Safety Framework
Amazon's framework establishing Critical Capability Thresholds across risk domains (CBRN, offensive cyber operations, and automated AI R&D) and the maximal-capability and safeguards evaluations required before deploying its frontier (Nova) models.
xAI Frontier Artificial Intelligence Framework
xAI's framework describing its risk categories (malicious use — including CBRN and cyber — and loss of control) with thresholds, benchmarks, risk-owner assignments, and governance procedures for its Grok models, published under the company's former name xAI.
Microsoft Frontier Governance Framework
Microsoft's framework for monitoring emerging frontier-model capabilities that could threaten national security or public safety, and for assessing and mitigating those risks before a model is deployed.
Frontier AI Safety Commitments, AI Seoul Summit 2024
Voluntary commitments made by AI companies and announced by the UK and Republic of Korea governments at the May 2024 AI Seoul Summit, under which the participating companies undertake to publish safety frameworks setting out how they identify and manage severe risks from their frontier models.
AI Incident & Harm Reporting Databases
Where AI incidents, harms and vulnerabilities are recorded. Counts drift daily, so none are quoted here — follow each link for the current figure.
AI Incident Database (AIID)
A public, searchable database that indexes real-world incidents in which AI systems caused or nearly caused harm.
AI Incidents and Hazards Monitor (AIM)
An OECD tool that documents AI incidents and hazards detected from international news media to build an evidence base for AI policy.
EU AI Act — reporting template for serious incidents involving general-purpose AI models with systemic risk (Article 55)
Official European Commission page providing the standardised template for reporting serious incidents involving general-purpose AI models with systemic risk under the EU AI Act.
MIT AI Incident Tracker
A dashboard that classifies incidents drawn from the AI Incident Database along risk, cause, harm and severity taxonomies.
AIAAIC Repository
An open, independent repository documenting incidents and controversies involving AI, algorithmic and automation systems.
AVID — AI Vulnerability Database
An open-source knowledge base of failure modes and vulnerability reports for general-purpose AI systems, mapped to risk taxonomies.
Adversarial Threat Taxonomies, Risk Frameworks & Safety Benchmarks
The named taxonomies, risk frameworks and safety benchmarks used to test and manage AI systems.
NIST AI Risk Management Framework (AI RMF 1.0 / NIST AI 100-1)
A voluntary framework, organized around the Govern, Map, Measure, and Manage functions, for identifying and managing risks across the AI lifecycle.
Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations (NIST AI 100-2e2025)
A report that defines a taxonomy and terminology of attacks and mitigations in adversarial machine learning, covering both predictive and generative AI systems.
MITRE ATLAS (Adversarial Threat Landscape for Artificial-Intelligence Systems)
A globally accessible knowledge base of adversary tactics, techniques, and real-world case studies against AI-enabled systems, modeled on MITRE ATT&CK.
OWASP Top 10 for LLM Applications (2025)
A community-developed list of the ten most critical security risks for large language model applications, each with a description and mitigation guidance.
MLCommons AILuminate
A safety benchmark from MLCommons that tests general-purpose chat systems against its published hazard taxonomy, with results reported per language.
How this switchboard is maintained
- Every entry names the authority that publishes it and links to that authority's own official page.
- Entries are described neutrally — what each resource is, not whether it is good. Nothing here is ranked, rated or recommended.
- Fast-moving details — framework versions, institute names, incident counts — are verified against the primary source and dated. Counts are never reproduced here because they change continuously; follow the link for the current figure.
- The AI Center links each organisation's own page and does not reproduce or re-originate its work.
Who coordinates AI safety and security evaluation internationally?
A network of national AI safety and security institutes coordinates technical research and evaluation of advanced-AI risks across governments.
The International Network of AI Safety Institutes was launched in San Francisco on 21 November 2024 by founding members including Australia, Canada, the European Commission, France, Japan, Kenya, the Republic of Korea, Singapore, the United Kingdom and the United States; its mission statement is published by the Government of Canada. Several members have since been renamed — the UK's is now the AI Security Institute, and the United States' is the Center for AI Standards and Innovation (CAISI) at NIST.
Source: International Network of AI Safety Institutes — Mission statement (Government of Canada / ISED) ↗
What is a frontier AI safety framework?
A frontier AI safety framework is a policy a developer publishes describing the dangerous-capability thresholds it will watch for and the safeguards it commits to apply before training or releasing models that reach them.
These frameworks are published by the developers themselves — for example Anthropic's Responsible Scaling Policy, OpenAI's Preparedness Framework and Google DeepMind's Frontier Safety Framework — under the Frontier AI Safety Commitments made by companies at the AI Seoul Summit in May 2024. The AI Center lists each developer's own current framework and links to it; it does not compare or rate them.
Source: Frontier AI Safety Commitments, AI Seoul Summit 2024 (GOV.UK) ↗
Cite this directory
1BusinessWorld AI Center, "AI Safety, Alignment & Security." https://1businessworld.com/ai-center/ai-safety-and-security/ Version as of July 28, 2026.
The AI Center is informational only. It is provided by 1BusinessWorld strictly for general informational and educational purposes. Nothing in the AI Center constitutes, or should be construed as, legal, regulatory, compliance, technical, engineering, security, investment, financial, or other professional advice, or a recommendation, endorsement, solicitation, or offer regarding any technology, product, model, provider, framework, or course of action. 1BusinessWorld is not a law firm, regulatory authority, standards body, conformity-assessment or certification body, or investment adviser, and nothing in the AI Center creates any advisory, fiduciary, attorney-client, or other professional relationship with 1BusinessWorld. Although the AI Center references official materials published by legislatures, regulators, standards bodies, research organizations, and other named authorities, 1BusinessWorld makes no representation or warranty, express or implied, as to the accuracy, completeness, timeliness, or fitness for any purpose of any content, and, to the fullest extent permitted by law, disclaims all liability for any loss or damage of any kind arising directly or indirectly from the use of, or reliance on, any information presented. Laws, regulations, standards, technical practices, and AI capabilities change frequently and differ by jurisdiction; readers must verify all information against the current official text or source and consult qualified legal, compliance, technical, and other professional advisors before acting. Any decision relating to the development, deployment, procurement, or governance of AI systems is made solely at the reader's own risk. Last reviewed: July 28, 2026.