Primary source library

Read what our guides are built on.

No source is treated as infallible. Industry frameworks describe what those companies say they do; standards describe recommended practices; synthesis reports summarize the state of evidence and disagreement.

International synthesis

International AI Safety Report 2026

A multi-expert synthesis led by Yoshua Bengio covering capabilities, misuse, reliability, loss of control, and risk-management evidence.

Open primary source ↗

Standards and public risk management

NIST AI Risk Management Framework

U.S. National Institute of Standards and Technology framework for managing AI risks across design, deployment, and use.

Open primary source ↗NIST Generative AI Profile

Cross-sector companion resource focused on generative AI risks and mitigations.

Open primary source ↗

Frontier-lab safety frameworks

OpenAI Preparedness Framework

OpenAI’s public framework for tracking capabilities that could create severe harm and the safeguards tied to them.

Open primary source ↗Anthropic Responsible Scaling Policy v3

Anthropic’s voluntary framework for catastrophic-risk capability thresholds, safeguards, and governance.

Open primary source ↗Google DeepMind Frontier Safety Framework

DeepMind’s framework for identifying critical capabilities and applying mitigation protocols.

Open primary source ↗

Search and publishing standards used by this site

Google Search: helpful, reliable, people-first content

The quality baseline used to avoid search-engine-first mass publishing.

Open primary source ↗Google Search: JavaScript SEO basics

Why this React/Vite site prerenders public pages while preserving normal client navigation.

Open primary source ↗Google Images SEO best practices

Guidance informing crawlable, descriptive, high-resolution article imagery.

Open primary source ↗