This document provides guidance on using RALIA responsibly, including its AI-powered features. By using RALIA, you agree to follow these guidelines and our Terms of Service.
1. Understanding RALIA
What RALIA Is
- AI-powered compliance management platform
- Tool for navigating AI Act requirements
- Draft documentation generator
- Regulatory monitoring service
- Workflow management system
What RALIA Is Not
- Not a legal advice provider
- Not a replacement for professionals
- Not a compliance guarantee
- Not an official certification
- Cannot make binding legal decisions
Your Role as Deployer
Under the EU AI Act, organisations using RALIA are considered "deployers" of an AI system. You have responsibilities for how AI outputs are used within your organisation. RALIA supports you, but ultimate decisions and accountability remain with your organisation.
2. Intended Use
2.1 Approved Use Cases
Compliance Self-Assessment
Internal AI system assessments against regulatory frameworks
Documentation Drafting
Generate draft compliance documents for review
Regulatory Research
Understand requirements and their application
Compliance Monitoring
Track regulatory developments
Team Collaboration
Coordinate compliance activities
2.2 Prohibited Uses
Providing Legal Advice
Using outputs as substitute for legal counsel
Misrepresenting Compliance
Claiming certification based on RALIA
Automated Decisions
No human review of AI decisions
Malicious Use
Manipulating or exploiting AI systems
Fraudulent Activity
Creating false documentation
3. Working with AI Outputs
3.1 Review Before Use
All AI-generated content should be reviewed before use. Use this checklist:
Review Checklist
- Check accuracy against official sources
- Verify applicability to your situation
- Ensure content is current and up-to-date
- Adapt generic guidance to your context
- Have qualified personnel review critical content
3.2 Understanding Limitations
AI Output Limitations
- Accuracy:AI may produce errors or hallucinations
- Completeness:Outputs may not cover all relevant factors
- Currency:AI training has a knowledge cutoff date
- Context:AI may not understand your specific circumstances
- Jurisdiction:Guidance may not account for local variations
3.3 Citation and Attribution
Do:
- Review AI outputs before presenting them
- Verify citations to source documents
- Check original sources for critical info
- Note when content is AI-assisted
Don't:
- Present AI outputs as original analysis without review
- Assume AI summaries capture all details
- Rely solely on AI citations without verification
- Skip professional review for important decisions
4. Data Input Guidelines
Appropriate Data
The following data is appropriate to input:
- Descriptions of your AI systems and use cases
- Compliance-related questions and scenarios
- Organisational policies and procedures for review
- Technical documentation about AI implementations
- Questions about regulatory requirements
Avoid Inputting
Avoid inputting the following sensitive data:
- Personal data of individuals (names, contact details)
- Special category data (health, biometric, political)
- Trade secrets or highly confidential information
- Security credentials, passwords, or API keys
- Information subject to legal privilege
- Data you lack authority to process
Data Quality Matters
The quality of AI outputs depends on the quality of inputs. Provide accurate, complete information; clear, specific questions; relevant context; and corrections when AI makes incorrect assumptions.
5. Security Responsibilities
5.1 Account Security
Security Checklist
- Use strong, unique passwords
- Enable two-factor authentication
- Never share login credentials
- Report suspected unauthorised access
- Log out on shared devices
5.2 Data Handling
Do:
- Store exported documents securely
- Share only with authorised individuals
- Apply appropriate access controls
- Delete data when no longer needed
Don't:
- Store sensitive exports unencrypted
- Share access broadly without need
- Leave exports accessible to unauthorised users
- Retain data beyond retention periods
5.3 Incident Reporting
Report any security concerns or suspicious activity immediately:
security@risqbase.com6. Organisational Use
6.1 Team Responsibilities
- 1Ensure team members understand these guidelines
- 2Assign appropriate roles and permissions
- 3Establish internal review processes for AI outputs
- 4Maintain audit trails of compliance decisions
- 5Designate personnel responsible for AI governance
6.2 Integration with Compliance Programmes
RALIA should complement, not replace, your compliance programme. Integrate outputs into existing workflows, document how AI was used, ensure authorised personnel give final approvals, and keep records of modifications to AI-generated content.
6.3 Regulatory Interactions
Do:
- Be transparent about AI tool usage
- Explain your human oversight processes
- Maintain records of human review
- Document your AI governance approach
Don't:
- Claim RALIA outputs as certifications
- Hide AI involvement from regulators
- Suggest automated compliance without oversight
- Present AI drafts as final submissions
7. Feedback & Reporting
7.1 What to Report
7.2 How to Report
General Issues
support@risqbase.com
AI Concerns
ai-feedback@risqbase.com
Security Issues
security@risqbase.com
Abuse Reports
security@risqbase.com
8. Compliance with Guidelines
Enforcement
Violation of these guidelines may result in:
- • Warning and requirement to cease prohibited behaviour
- • Temporary suspension of account access
- • Permanent termination of service
- • Reporting to relevant authorities where required by law
Updates to Guidelines
We may update these guidelines to reflect changes in best practices, regulations, or platform capabilities. Material changes will be communicated to users.
9. Additional Resources
Responsible AI
How we design and govern AI systems
Privacy Policy
How we handle your data
Terms of Service
Legal terms governing platform use
10. Contact
Questions about these guidelines or responsible use of RALIA? Contact us:
Post
RisqBase d.o.o.
Zagreb, Croatia
Document History
| Version | Date | Summary |
|---|---|---|
| 2.0 | January 2026 | Enhanced design with checklists and improved navigation |
| 1.0 | January 2026 | Initial publication |