Describe an AI system. Get the provisions that bind — deterministically.
Measured,
not modelled.
417 frozen statutory provisions. Four lenses. Zero models in the verdict path — every verdict cited and signed, every failure published.
This is not an AI system. It is a deterministic rule-based tool.
The console pattern-matches your question against frozen statute text and renders the matching provision. There is no inference call in the console's code, and nothing you type into it is transmitted or stored. The page also carries an email form posting to /api/subscribe, and — like every page — the global Sovereign dock, which IS an AI-system surface (see the note at the top of this file).
Rule-based systems defined solely by natural persons fall outside the Article 3(1) definition (Recital 12), so Article 50(1) does not apply. We say so rather than leave it unsaid. Every surface and its classification
Describe an AI system. Four lenses read it against 417 frozen statutory provisions and tell you which bind — deterministically, with the provision cited.
THE INSTRUMENTS
Measured, not modelled.
417 statutory provisions · 193 deterministic GovBench items · every verdict a predicate an auditor can recompute.
CSOAI · measurement
The measurement lens
The UK runs a principles-based, regulator-led approach (no single AI Act yet); the AI Safety Institute leads frontier evaluation.
Built for the people who get audited
Governance shouldn't cost more than the AI it governs. Open-source core · free to start · own your data · no per-seat rent.
Honest by design: we show what's true and verifiable, not badges we don't hold. Formal certifications (e.g. SOC 2, ISO 42001) are pursued as the platform matures.
Zero AI Safety SolutionsUntil Now.
For years, the AI safety landscape was fragmented and ineffective. No unified system existed to train analysts, monitor systems, or protect workers.
No Jobs for AI Analysts
There was no structured profession or marketplace for AI safety monitoring
CSOAI creates thousands of remote, high-paying AI Safety Analyst positions
No Human Oversight at Scale
Governments can't monitor millions of AI systems with limited staff
Certified analysts + the 33-seat Council design (measured status on the public ledger)
No Unified Compliance
7+ frameworks, each requiring separate compliance efforts
One registration = compliance scores for ALL frameworks automatically
No Economic Safety Net
AI displaces workers but no mechanism shares AI's economic gains
Prosperity Fund provides UBI when AI-caused unemployment rises
Choose Your Path
CSOAI has something for everyone. Find your place in the AI safety ecosystem.
How it works, in one glance
Many frameworks. One signed proof.
Governance rules from every jurisdiction reach a fault-aware consensus council consensus — and out comes one Ed25519-signed record anyone can verify offline, without trusting a vendor dashboard.
Why AI Safety Matters Right Now
Every day, AI systems make decisions affecting millions of lives—often without proper oversight. These aren't hypothetical risks. They're happening today.
Hiring Discrimination
AI hiring tools have been found to disadvantage women — Amazon scrapped a recruiting model that penalised CVs mentioning “women’s” (Reuters, 2018)
Careers shaped by unaudited models
Medical Errors
AI diagnostic tools can underperform for under-represented groups when trained on non-diverse data (peer-reviewed literature)
Lives put at risk by biased training data
Financial Bias
AI lending and pricing models have produced discriminatory outcomes, drawing regulator attention (e.g. US CFPB, UK FCA)
Families denied loans, homes, opportunities
Autonomous Failures
Autonomous-vehicle incidents remain under active safety investigation by regulators (e.g. NHTSA)
Safety gaps in real-world deployment
Enter the OS
Five doors into the Sovereign OS
Build your identity, connect your world, see governance on the globe, choose your consensus, and browse every regulation - all live.
2 AUG 2026EU AI Act transparency + GPAI go live - run your free signed risk check ->One Platform. Everything Connected.
See how CSOAI's systems work together — from global frameworks to local enforcement, from training to employment, from monitoring to prosperity.
CSOAI Ecosystem
Interactive visualization of integrated AI safety systems
Ecosystem Legend
Hover to explore connections
The CSOAI Advantage
Unity from fragmentation. One operational platform integrating all global frameworks.
The Problem: Fragmentation
Today's AI governance landscape is chaos
The Solution: CSOAI
One unified platform for all AI safety needs
The Only Platform Solving All Four Problems
We're not just a tool. We're building the infrastructure for safe AI governance—and creating jobs while we do it.
For Every Stakeholder
Whether you're a citizen, enterprise, government, or aspiring analyst — CSOAI has something transformative for you.
Our 7 Frameworks
Master every major AI governance framework. One platform, complete global coverage.
AI Safety Governance Network
Explore the global landscape of AI safety stakeholders, regulations, and frameworks. CSOAI integrates with all major initiatives to provide unified oversight.
European Commission
Europe
UK DSIT
Europe
US NIST
Americas
China CAC
Asia
Singapore IMDA
Asia
Canada AIDA
Americas
Japan METI
Asia
South Korea MSIT
Asia
Australia DTA
Asia
Brazil ANPD
Americas
India MeitY
Asia
UAE AI Office
Asia
OECD
Global
UNESCO
Global
ISO/IEC
Global
IEEE
Global
Partnership on AI
Global
World Economic Forum
Global
UN AI Advisory Body
Global
G7 AI Process
Global
GPAI
Global
US AI Safety Institute
Americas
MIT CSAIL
Americas
Stanford HAI
Americas
Oxford FHI
Europe
Cambridge CSER
Europe
Tsinghua AI Institute
Asia
Berkeley CHAI
Americas
Alan Turing Institute
Europe
Mila Quebec
Americas
ETH Zürich AI Center
Europe
KAIST AI
Asia
Anthropic
Americas
Google DeepMind
Europe
OpenAI
Americas
Meta AI
Americas
Microsoft RAI
Americas
IBM AI Ethics
Americas
Baidu AI
Asia
NVIDIA Trustworthy AI
Americas
EU AI Act
Europe
NIST AI RMF
Americas
ISO/IEC 42001
Global
China TC260
Asia
UK AI Safety Framework
Europe
Singapore Model Framework
Asia
IEEE 7000 Series
Global
SOAI-PDCA
Global
CSOAIThe Maternal Covenant
AI should want to protect humans the way a mother wants to protect a child.
— The Maternal Covenant principle
Control Paradigm (Fails)
- •Adversarial: AI vs Humans
- •Arms race: ever-stronger constraints
- •Inevitable failure with superintelligence
Covenant Paradigm (Works)
- •Partnership: AI + Humans together
- •Cooperative: mutual benefit design
- •Sustainable: scales with AI growth
Mother gives child allowance → AI gives humanity prosperity
The Byzantine Council
33 AI agents from multiple providers working together with fault-aware consensus. No single company can manipulate safety decisions.
23/33 Consensus Required
70% agreement for any safety decision. fault-aware consensus ensures reliability even if agents fail or are compromised.
Multi-Provider Diversity
Agents span GPT-4, Claude, Gemini, Llama, and more. No single AI company has majority control.
Human Oversight
Attested AI Safety Analysts review critical decisions. AI monitors AI, humans oversee all.
The Prosperity Fund
AI creates unprecedented wealth. The Prosperity Fund ensures everyone benefits.
Prosperity Fund Goal
Funded by AI company contributions (1-20% of revenues) - NOT by analyst fees
Tier 1 UBI
When AI-caused unemployment reaches 20%, first tier of Universal Basic Income activates.
Tier 2 UBI
When displacement reaches 40%, enhanced benefits activate to support affected workers.
Full UBI
At 70% displacement, full Universal Basic Income ensures everyone's economic security.
ALL 33 Courses 100% FREE
AI Safety Analyst is projected to be a top-10 profession by 2045. Get ahead now with completely free training - attestation exam £49, Analyst License £199/year.
Your Path to Success
Three steps to a new career
Train for Free
Complete ALL 33 courses on 7 frameworks. 100% online, self-paced, completely FREE.
Get Attested
Pass the attestation exam (£49 one-time). Your record is signed and independently verifiable.
Join the Roster
Get your Analyst License (£199/year) and join the analyst roster. Rates are set by the market and published as measured aggregates once live.
CSOAI Attestation
AI System Compliance Badge
Licensing & Compliance
We license AI systems for safety compliance. Ongoing monitoring. Assessment against all frameworks. Global recognition.
Built on Transparency
"We built CSOAI on a simple principle: measure what is real, sign what we publish, and never claim what we cannot show. Governance that works is governance you can audit."
One OS, every role in AI governance
From compliance officers to the public — here's what CSOAI does for each of you.
One platform, every framework
Instead of navigating seven regulatory frameworks separately, map each AI system once and get comprehensive, crosswalked coverage — sealed to Layer 0 for audit.
For compliance & risk teams
Learn it, then govern it
Free open-source training plus the AI Safety Analyst certification — build the skills, then run real assessments inside the OS.
For practitioners
Accountability by design
A 33-seat Council of AI drawn from different providers is designed so no single company can game the outcome — accountability that's structural, not promised.
For policy & research
Report, and see action
The public Watchdog lets anyone flag an AI incident and watch the transparency dashboard — real follow-through, not corporate promises.
For the public
Partnership, not adversarial control
CSOAI is designed as partnership between people and AI — an approach built to scale with AI capability instead of fighting it.
For governance leaders
Provable when auditors ask
Every decision is Ed25519-signed to Layer 0 and verifiable offline. Comply once, and it crosswalks everywhere.
For teams under deadline
Everything You Need to Get Started
Free tools, guides, and resources to help your organization navigate AI governance with confidence.
AI Governance Readiness Assessment
Take our free 5-minute assessment to evaluate your organization's AI governance maturity and get personalized recommendations.
Global Regulation Tracker
Track AI governance frameworks across 40+ nations in real-time. Filter by region, status, and framework type.
AI Glossary
60+ AI governance terms defined clearly. From algorithmic accountability to zero-shot learning — everything you need to know.
Industry Solutions
Tailored governance frameworks for Financial Services, Healthcare, Government, Technology, and 4 more sectors.
Case Studies
See how organizations worldwide are implementing AI governance with CSOAI. Real results, real impact.
Trust Center
Our security certifications, compliance frameworks, data protection measures, and infrastructure details.
Join the Movement
The future of AI safety starts with you. Choose your path and become part of humanity's unified response to AI.
Start Training
Free courses, attestation, career
Enterprise
Register systems, ensure compliance
Founding Member
Shape the future, exclusive benefits
Questions? Contact us at [email protected]
The measurement stack, by the numbers
source: corpus-watch baseline + flywheel + AIR-Bench harvester, 2026-08-01
Frequently asked questions
Straight answers about what CSOAI measures — and what it does not claim.
What does CSOAI actually measure?
AI system behaviour against published regulatory provisions, across four axes: Governance, Safety, Provenance and Continuity (GSPC). Every published number traces to a signed measurement artefact — not to a self-assessment form.
Is CSOAI a certification body?
No. CSOAI is a measurement body. You receive a signed, provision-anchored attestation record of measured behaviour. Accredited conformity certification requires a national accreditation chain, and we hold no such accreditation — we will not pretend otherwise.
How current is the regulatory data behind the scores?
Living, not frozen. Our public corpus-watch instrument re-hashes 127 provisions across four instruments (EU AI Act, UK GDPR, DPA 2018, NIS2-UK) every day and publishes drift events. When the law moves, measurements anchored to it are flagged for re-review.
What happens when something cannot be measured?
It is reported as UNMEASURED — never as a pass and never as a zero. An unreachable lane, a missing model, or a coverage gap is shown honestly with its n and its limits.