AI4Growth Ethical, Compliant and Trustworthy AI

Omina Technologies > Conference > Webinar > AI4Growth Ethical, Compliant and Trustworthy AI

AI4Growth Ethical, Compliant and Trustworthy AI

  • Posted by: MartinCanter
AI4Growth

Key Take-Aways

Omina Technologies was invited as a partner to the AI4Growth partner session on ethical, compliant and trustworthy AI.

Thanks to Magali Feys for a great overview of the legislation and guidelines for #ethical, #compliant and #trustworthyai!

  • The Belgian extra-contractual law (article 1382, Civil Code) is too limited. As the harm of AI is typically manifested later, proving the cause-effect relation is difficult. It is advised to add additional contract clauses.
  • Ethical Guidelines for Trustworthy AI: Might not be enough as ethical principles are interpreted differently depending on the culture.
  • Important to complement ethical guidelines with an AI risk assessment: test environment, risk assessment, etc.
  • Legislation differences: US: federal and state law next to each other, more case-based, Algorithmic Accountability Act has not yet been approved, China: state-controlled AI, regulations by CAAI
  • Intellectual property and AI:
    • copywright on object and source code and human intervention is required
    • database IP: database structure can be protected, database content too but it should not be directly related to the company’s main activities
  • Data Protection and Privacy: GDPR: only for personal identifiable data, do not use anonymization but use use-case specific pseudonymization as this reduces the chance to identify a person indirectly