A high-risk AI system is a system whose use can significantly affect people's rights, safety, or life opportunities. This category includes, for example, the use of AI in selecting job applicants and evaluating employees, in assessing creditworthiness, in access to education and essential services, in biometric identification, or in critical infrastructure. What matters is the purpose of use, so a commonly available tool becomes high-risk once deployed in such a process, even if the company doesn't realise it. This category comes with the strictest package of obligations: risk management, data quality requirements, technical documentation, record-keeping, human oversight, accuracy and robustness, and informing the data subjects. If you are considering such a use, assess it in advance, not after deployment.
See also: AI risk classification, Fundamental rights impact assessment, Human oversight of AI.
