[ The Standard ]
The standard for AI the world can trust.
Safety Critical Labs is an independent certification authority for AI in high stakes systems. We assess it against the failures it must prevent, and hold it to the rigor of human spaceflight.
Safety Critical Labs
AI Certification Authority
Certificate of Conformance
This is to certify that
System Name
Organization · Version X.X · Assessed Month YYYY
has been formally assessed against the
SCL AI Requirements Framework
and certified at the Safety Critical level for the operational claim stated below.
Operational Claim
One sentence operational claim describing the certified role and intended scope of the system.
Certificate No.
SCL-CERT-2025-001
Issued
DD Mon YYYY
Framework DOI
10.5281/zenodo.21924937
Principal
Safety Critical Labs
Independent Certification · Safety Critical Labs
Safety Critical Labs
Conformance Verification
Assessed Requirement Areas
AI-1Operational and data foundations
AI-2Addressing AI bias
AI-3ML test coverage
AI-4Continuous validation
AI-5Hallucination prevention
AI-6Out of distribution detection
AI-7Adversarial robustness
AI-8Explainability
AI-9Human and AI teaming
AI-10Privacy and data protection
AI-11Multi-model systems
AI-12Neural network requirements
AI-13Continuous learning and adaptation
Evaluated across all thirteen areas of the SCL AI Requirements Framework. Scan to confirm this record against the public register.
Independent Certification · Safety Critical Labs
Certificate of Conformance
The failure it prevents
View in the framework →
The thirteen requirement areas of the framework. Each exists to prevent a specific failure. Select one.
The failure it prevents
View in the framework →
Thirteen requirement areas
Ask SCL
Ask SCL about certification, the framework, and the process.
Readiness assessment
How ready is your AI system to certify? A free self assessment across the thirteen requirement areas.
[ Regulatory ]
Regulatory activity from the Federal Register, EU Commission, and NIST, refreshed throughout the week.
Agentic Misalignment: How LLMs could be insider threats
Anthropic
Taking a responsible path to AGI
Google DeepMind
AI Index 2025: State of AI in 10 Charts
Stanford HAI
Alignment Faking in Large Language Models
Alignment Forum
Explained: Generative AI's environmental impact
MIT News
MIT researchers release a repository of AI risks
TechCrunch