AI Liability Shield™ — Client Portal

Loading your compliance data...

—

Risk Score

AI Compliance Status

Annex IV — Model Version Control

Detected Models

Register Model

Annex IV — Technical Documentation

Technical Documentation

EU AI Act 2024/1689 — Annex IV

Document Status

DRAFT

1. General Description of the AI System

System Name

Provider — legal entity name · Annex IV 1

Sector / industry · Annex IV 1

Version

Intended Purpose

Risk Category

Annex III category (high-risk only)

Hardware the system runs on - Annex IV 1(e)

Software / firmware version and update requirements - Annex IV 1(c)

Market placement forms - Annex IV 1(d)

Interaction with external hardware / software / other AI systems - Annex IV 1(b)

User interface and instructions for use for the deployer - Annex IV 1(g)(h)

2. AI Model & Training Data

Model Architecture

Model Provider

Training Data Description

Development methods and third-party pre-trained models used - Annex IV 2(a)

Design specifications: general logic, optimisation objective, key assumptions, expected output and quality, trade-offs - Annex IV 2(b)

Computational resources - training and validation - Annex IV 2(c)

Data provenance, scope, labelling and cleaning - Annex IV 2(d)

Robustness and bias metrics; reference to dated and signed test report - Annex IV 2(g)

3. Performance, Accuracy and Operational Metrics

—

Total Requests

—

Avg Latency ms

—

Error Rate %

—

Active Users

Accuracy - overall and per affected group/person - Annex IV 3

Foreseeable unintended outcomes and risks (health/safety, fundamental rights, discrimination) - Annex IV 3

Input data specifications - Annex IV 3

Appropriateness of the performance metrics for this system - Annex IV 4

4. Testing & Validation

Testing Methodology

Last Test Date

Known Limitations

Validation / testing data used and characteristics - Annex IV 2(g)

Responsible person and sign-off for testing - Annex IV 2(g)

5. Version Control & Change Log

Pre-determined changes and continuous-compliance measures (where applicable) - Annex IV 2(f)

6. Architecture Documentation AI GENERATED

Click Generate to auto-generate.

7. Intended Purpose AI GENERATED

Click Generate to auto-generate.

8. Limitations AI GENERATED

Click Generate to auto-generate.

FRIA — Fundamental Rights Impact Assessment

0/6

SECTIONS COMPLETED

FRIA SCORE

—

Чл. 6 — Класификация на риска (EU AI Act)

Текуща класификация

—

Все още не е класифицирана.

Случай на употреба по Приложение III

Чл. 5 — забранена практика (манипулиране, експлоатиране на уязвимост, разпознаване на емоции на работното място/в образованието и др.) Системата извършва профилиране на физически лица Дерогация по чл. 6(3) — тясна процедурна/подготвителна задача (само ако НЯМА профилиране) Прилагат се задължения за прозрачност по чл. 50 Висок риск на друго основание (напр. защитен продукт по Приложение I)

Правна преценка по чл. 6. Потвърдете и запишете — тази стойност захранва документите по Приложение IV.

Art. 12 — Log Retention Policy

Retention Status

Policy Rules

Art. 61 — Post-Market Monitoring

Monitoring Timeline

Trend Analysis

Next Scheduled Review

—

Weekly Review

—

Monthly Report

—

Annual Assessment

Art. 15 — Security Monitoring

—

Prompt Injections

Scanning...

—

Jailbreak Attempts

Scanning...

—

Unauthorized Access

Scanning...

—

Security Score

Analyzing...

Threat Detection Log

Policy Engine — Custom Rules

Add New Rule

IF Condition

THEN Action

Severity

Active Rules 0

DEFAULT PLATFORM RULES (always active)

IF risk_level = critical → Block + Log Incident + Alert (Art. 15)

IF policy_compliant = false → Log + Flag for Review (Art. 9)

IF is_incident = true → Auto-report to Art. 72 log (Art. 72)

Art. 14 — Human Oversight

System Status

System Active

All AI operations running normally

Oversight Controls

ℹ Спирането прекъсва реално AI повикванията само за системи, свързани през Shield API Proxy. При само пасивен мониторинг (tracker.js) действието се записва в одита, но техническото прекъсване трябва да се извърши във вашата среда.

Art. 14 — Human Decision Override

Recent AI Decisions — Pending Review

🔍 Investigation Workflow

Инциденти под разследване — анализ на първопричините (RCA), времева линия и експорт за регулатор (Чл. 72).

Art. 72 — Incident Reporting

Reporting Status

Regulatory Contacts

BULGARIA

Commission for Personal Data Protection

kzld.bg · +359 2 915 3580

EU LEVEL

European AI Office

digital-strategy.ec.europa.eu

Чл. 3 — Роля на субекта

ⓘ КУМУЛАТИВНИ РОЛИ
Един субект може да изпълнява повече от една роля едновременно (напр. създава GPAI модел и пуска конкретна високорискова система). В такъв случай се прилагат всички приложими набори задължения. Прегледайте всеки релевантен раздел поотделно. Чл. 25 урежда отговорностите по веригата на стойността.

Art. 26 — Implementer Obligations Checklist

0/8

OBLIGATIONS COMPLETED

Complete all obligations to achieve Art. 26 compliance

Risk Scoring Matrix — Art. 9

—

RISK SCORE

—

Risk Dimensions

Probability × Impact Matrix

Probability ↓ / Impact →	1 — Minimal	2 — Low	3 — Medium	4 — High	5 — Critical

Identified Risks Register

—

Total AI Requests

✓ Logged

—

Avg Latency (ms)

Monitored

—

Error Rate (%)

Tracked

—

Drift Score

Analyzed

Recent AI Requests

Time	Model	Status	Latency	Risk	Prompt	Response
Loading...

Legal Defense Score

—/100

Audit Trail—/25

Incident Response—/25

Regulatory Mapping—/25

Evidence Archive—/25

AI Behavioral Fingerprint

DOMINANT MODEL

—

TOKEN RATIO

—

SYSTEM STABILITY

—

Compliance Documents

📄

Monthly Report

EU AI Act Article 72 compliance report with full metrics and audit trail summary.

⚖️

EU Declaration

Formal EU Declaration of Conformity for regulatory submission. Art. 72 & 73 compliant.

🛡️

Insurance Package

Risk quantification report for insurance underwriting. Includes risk score and compliance status.

EAA Accessibility Scanner

Scan URL

Scan History

Incident Register — Art. 72

TOTAL

OPEN

INVESTIGATING

RESOLVED

CRITICAL OPEN

Report New Incident

Recent Incidents

AI Workforce Intelligence

👥

Active Users

📈

Requests This Week

⚠️

Anomalies

🏆

Top User

Installation Guide

📊

JS Tracker — Site Monitoring

Covers: page views, JS errors, load time, EAA scanning. Add one line to your website.

STEP 1 — Add to <head> of your website

<script src="https://shield.gistoianova.com/tracker.js?id=CLIENT_ID"></script>

WORDPRESS — Appearance → Theme Editor → header.php

Or use "Insert Headers and Footers" plugin

✓ Page views ✓ JS errors ✓ Load time ✓ EAA auto-scan

🔌

API Proxy — AI Monitoring (No Docker)

Covers: AI requests, tokens, latency, compliance reports. Change one URL in your app settings.

STEP 1 — Replace OpenAI URL in your app

shield.gistoianova.com/ai/CLIENT_ID/v1/chat/completions

STEP 2 — Keep your API key unchanged

Authorization: Bearer YOUR_OPENAI_KEY (unchanged)

✓ AI requests ✓ Tokens ✓ Latency ✓ PDF reports

🐳

Docker Agent — Full Monitoring

Covers: everything above + behavioral fingerprint, workforce intelligence, drift detection.

STEP 1 — Run in terminal (requires Docker)

docker run -d -e CLIENT_ID=CLIENT_ID -e AILS_ENDPOINT=https://shield.gistoianova.com/api/ingest -e OPENAI_API_KEY=YOUR_KEY -p 8080:8080 galinass/ails-agent:latest

STEP 2 — Point your app to local agent

localhost:8080/v1/chat/completions

✓ Full AI monitoring ✓ Workforce ✓ Fingerprint ✓ Drift

⭐

JS + Docker = 100% Coverage

Combine JS Tracker + Docker Agent. Complete monitoring with no blind spots.

STEP 1 — Add to website <head>

<script src="https://shield.gistoianova.com/tracker.js?id=CLIENT_ID"></script>

STEP 2 — Run Docker Agent

docker run -d -e CLIENT_ID=CLIENT_ID -p 8080:8080 galinass/ails-agent:latest

✓ Page views ✓ AI requests ✓ EAA ✓ Behavioral fingerprint

💳 Billing & Subscription

Trial Plan

Loading...

€0

per month

TRIAL ENDS

—

NEXT BILLING

—

STATUS

—

STARTER

€499/mo

For small teams getting started with AI compliance

✓ Up to 10,000 AI requests/mo

✓ Basic compliance reports

✓ EU AI Act monitoring

✓ Email support

✗ FRIA & Annex IV

✗ Workforce Intelligence

RECOMMENDED

PROFESSIONAL

€1,200/mo

Full EU AI Act compliance for growing businesses

✓ Unlimited AI requests

✓ Full compliance suite

✓ FRIA & Annex IV PDF

✓ Workforce Intelligence

✓ Incident Management

✓ Priority support

ENTERPRISE

Custom

For large organizations with complex AI deployments

✓ Everything in Professional

✓ Custom integrations

✓ Dedicated compliance officer

✓ SLA guarantee

✓ On-premise deployment

✓ 24/7 support

🔒

Secure payments powered by Stripe

All transactions are encrypted and PSD2/SCA compliant. VAT automatically applied for EU businesses.