Ethics and privacy in AI regulation: Navigating challenges and strategies for compliance

A new summer of artificial intelligence (AI) started a year ago, promising tantalising technical development and efficiencies of scale, while in parallel the Internet is flooded with advice, notes and analysis of AI’s impact and risks. Although the potential use of AI is promising and could help solve very real human challenges, the risks and societal impact are real too. With AI infiltrating all areas of life, such as online platforms, work, healthcare, social services and the justice system, it is essential that it is developed within key safety parameters. Furthermore, it is no secret that for AI to be effective it needs to process vast quantity of data, which is at odds with the General Data Protection Regulation (GDPR) principles of data minimisation. Businesses are repeatedly told to mitigate such risks on fundamental rights, privacy, discrimination, biases, etc. with stringent privacy and AI governance, all within an ethical framework and in compliance with existing legislation. Among the bombardment of information, this paper seeks to provide practical guidelines to comply with existing privacy regulation while implementing safe and trustworthy AI. The first part considers compliance with the GDPR while developing or using AI, while the second part provides practical recommendations in relation to the implementation of an ethical AI framework.