In order to ensure a high level of trustworthiness of high-risk AI systems, the manufacturer or its authorised representative should take appropriate corrective action where necessary.