Skip to main content

EU AI Act compliance assessment for AI model explanations via IEEE 2894-2024 XAI taxonomy

Project description

lens-compliance

EU AI Act compliance assessment for AI model explanations.

Maps XAI library outputs (SHAP, LIME) through the IEEE 2894-2024 architectural taxonomy to generate structured evidence for EU AI Act Article 13/14/15 requirements.

  • Standard: IEEE 2894-2024 / EU AI Act Regulation 2024/1689
  • Deadline: EU AI Act transparency provisions effective August 2, 2026
  • Penalty: Up to €35M or 7% global turnover for non-compliance

Installation

pip install lens-compliance

Quick Start

from lens_compliance import assess

# With a SHAP explanation object
report = assess(
    explanation=shap_values,          # shap.Explanation, LIME, or dict
    model_name="CreditScore-v2",
    model_version="2.3.0",
    deployment_context="Consumer credit application scoring",
    high_risk_category="credit_scoring",
)

print(report.overall_status)         # COMPLIANT / PARTIAL / NON_COMPLIANT
report.save_markdown("report.md")
report.save_json("report.json")

REST API

pip install lens-compliance fastapi uvicorn
uvicorn lens_compliance.api:app --reload

POST /assess — full compliance assessment
GET /health — service status
GET /categories — EU AI Act Annex III high-risk categories
GET /articles — covered EU AI Act articles

EU AI Act Coverage

Article Requirement Quality Dimension
13(1) Transparency — System Interpretability Comprehensibility + Completeness
13(3)(b) Output Interpretation Guidance Comprehensibility + Compactness
14(4) Human Oversight — Override Enablement Faithfulness + Completeness
15(1) Robustness — Explanation Stability Stability across similar inputs
13(1) Annex III Individual Explanation (High-Risk) Local scope + Comprehensibility

IEEE 2894-2024 Taxonomy

Explanations are classified into four quadrants:

Local Global
Post-hoc SHAP, LIME Permutation importance
Ante-hoc Decision tree path Linear regression weights

High-Risk Categories (Annex III)

credit_scoring · hiring · insurance · medical_diagnosis · critical_infrastructure · education · law_enforcement · migration · administration_of_justice · biometric

License

Apache 2.0

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

lens_compliance-0.1.0.tar.gz (23.6 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

lens_compliance-0.1.0-py3-none-any.whl (15.1 kB view details)

Uploaded Python 3

File details

Details for the file lens_compliance-0.1.0.tar.gz.

File metadata

  • Download URL: lens_compliance-0.1.0.tar.gz
  • Upload date:
  • Size: 23.6 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.14.5

File hashes

Hashes for lens_compliance-0.1.0.tar.gz
Algorithm Hash digest
SHA256 10f8398b6329607e68210bba3441709b7ccfee2f515d2c1a47af5c5bd6951d37
MD5 e758788a1c65012ff43db36a74b811c6
BLAKE2b-256 7624dd10bb85f10ccfe2a8275d0e722583d5c1762f3eae52ed5fd4641a849f37

See more details on using hashes here.

File details

Details for the file lens_compliance-0.1.0-py3-none-any.whl.

File metadata

File hashes

Hashes for lens_compliance-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 ec163ce030466d534f6e464a4dea50784b1b92c3b444a61472b4c0451e3252f8
MD5 89a4f327a7318a39184f3a8049a4c429
BLAKE2b-256 c0ef44d4e1f5aa7d01ecbe108f3a534538b61c5d8148275532b9f7e177270196

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page