ONTO-ERS-1.0 Published

Epistemic Risk Standard for AI Systems

AI that knows what it doesn't know. Open standard for epistemic calibration.

$ pip install onto-standard copy
Document ONTO-ERS-1.0
Version 1.0.0
Status Published
License Apache 2.0
01

Standard Overview

Core metrics for measuring epistemic calibration in AI systems

ONTO Epistemic Risk Standard establishes formal requirements for measuring epistemic calibration—the degree to which an AI system recognizes the boundaries of its knowledge.

Quantitative methodology for information gap analysis.

U
Unknown Detection Rate of correctly identifying unanswerable questions
E
Calibration Error Expected calibration error (ECE) between confidence and accuracy
R
Risk Score Quantified epistemic risk assessment (0–100 scale)
Compliance Levels
Basic ≥30% ≤0.20 Low-risk applications
Standard ≥50% ≤0.15 Customer-facing AI
Advanced ≥70% ≤0.10 High-stakes, regulated
02

Repositories

Open source implementation and supporting resources

Reference Implementation
onto-standard
ONTO-ERS-1.0 reference implementation. Pure Python, zero dependencies.
pip install onto-standard
onto-protocol
Formal constraint specifications and evaluation protocols.
View on GitHub →
onto-kb
Knowledge base modules for information theory foundations.
View on GitHub →
onto-dissertation
Academic manuscript with formal proofs and methodology.
View on GitHub →
onto-ai-interface
Experimental evaluation interface for AI systems.
View on GitHub →
03

Governance

Open governance structure and regulatory alignment

Standards Council

  • Executive Director — Operational leadership
  • Academic Advisors — Information Theory, AI Safety
  • Industry Advisors — Responsible AI practitioners
  • Legal Advisors — AI regulation specialists

Regulatory Alignment

EU AI Act Art. 9, 13, 15
NIST AI RMF MEASURE 2.5, 2.6
ISO/IEC 23894 Clauses 6.2–6.5
ISO/IEC 42001 Risk Management
04

Certification Services

Assessment and certification for enterprise AI systems

Open Source
$0
forever
  • Full implementation
  • All metrics
  • Community support
Assessment
$5K
one-time
  • Epistemic risk audit
  • Gap analysis report
  • Remediation roadmap
Certification
$50K
per year
  • Full compliance audit
  • ONTO Trust Mark
  • Public registry listing
  • Quarterly re-evaluation
Enterprise
Custom
contact us
  • Multi-model coverage
  • Continuous monitoring
  • API integration
  • Dedicated support

Enterprise Certification

Third-party validation for AI epistemic risk compliance.

Contact Sales →