The growing interest in Artificial Intelligence (AI) is linked to the expectation of perfection and responsiveness of these systems, but its diffusion is also associated with imperfections rooted in the nature of systems that reflect human biases. AI, based on Machine Learning and Deep Learning, can show ambiguities attributable to training data and the ‘dark side’ of certain behaviours, with different implications. Although these systems are applied in many contexts, some of them present critical issues that need to be seriously addressed due to the ethical, social and legal consequences they may produce. The intent of this investigation is to examine some known cases through their identification from web sources, in order to understand the causes and possible mitigation strategies among those proposed for appropriate use by reducing negative consequences. Every day the press covers significant cases, spotlights shining on major companies that have experimented, perhaps hastily, with applications with resounding failures. Episodes such as the failure of Microsoft’s Tay chatbot indicate the challenges in dealing with bias and discrimination. Even Syri, a Dutch anti-fraud system, was suspended for discrimination, highlighting the need to balance privacy and public interest in the digitisation of welfare. The case of discrimination by Amazon, and Joule’s ethical approach, underlines that eliminating bias requires continuous efforts. In the field of justice, the assessment of criminal causality entrusted to AI requires impartiality and faces challenges of design and privacy. Awareness of these issues is essential for an ethically responsible development of smart technologies (In this sense, the principles of the European Commission emphasise fairness, transparency and explainability in the use of AI, promoting a responsible and informed approach). However, only the continuous commitment of all those involved can guarantee safety and reliability. While science improves technologies, individual ethics must guide their evolution.